INTRODUCTION

Osteoporosis is a metabolic disease characterized by low bone mass and microstructural deterioration of the bone tissue that leads to increased bone fragility. The main complication involves the appearance of fragility fractures¹. Osteoporotic fractures are an important health problem² associated with high healthcare costs³. To prevent the appearance of fractures, different drugs are available that act on bone metabolism and are associated with reduced fracture risk⁴. The most commonly used in Spain are bisphosphonates⁵. However, in order to observe this protective effect, adequate therapeutic compliance is required⁶. In osteoporosis, as in all chronic diseases, compliance is low. In a recent study conducted in Spain, the overall persistence per year after commencing osteoporosis drug is 47%, and at two years, close to 27%⁷.

Therefore, correctly assessing therapeutic compliance is necessary in our consultations to ensure an adequate effect in reducing the risk of fracture. Classically, self-administered surveys have been used to assess therapeutic compliance, such as Morisky-Green and Haynes-Sackett questionnaires, although the latter tends to overestimate compliance⁸. In recent years, thanks to health system computerization, it is possible in certain cases to have access to drug dispensing data, so that the medication possession rate (MPR) can be calculated. This is used in many pharmaco-epidemiological studies^9-12, but not always available in day-to-day consultations.

Another possible way to assess compliance involves using bone remodeling markers, although there is little evidence in this regard and requires different determinations¹³. Determining carboxyterminal telopeptide of type I collagen (CTX) as a marker of resorption and of the amino terminal propeptide of type I procollagen (P1NP) as a formation marker is recommended¹⁴.

Our aim is to verify the usefulness of a single CTX determination to assess compliance in patients treated with bisphosphonates (the most prescribed drugs) for at least one year, in isolation or together with a classic therapeutic compliance questionnaire, such as Morisky- Green.

MATERIAL AND METHOD

RESULTS

Of the 115 patients selected, 15 were excluded for taking a drug other than an oral bisphosphonate (9, strontium ranelate and 6, raloxifene). The baseline characteristics are shown in table 1, with a higher proportion of densitometries prior to treatment and a lower proportion of patients treated with alendronate in the group of compliant patients. In 11 women, different errors were identified in the taking of the medication (an error in the taking of the medication -not on an empty stomach- an error in the medication taking method -with milk- and 10 errors in the waiting time of fasting). As in these 11 cases, the MPR was <0.8 and, therefore, they were considered non-compliant, and thus not excluded from the analysis.

Table 1. Baseline characteristics in the total number of patients and according to treatment compliance 

CaD: calcium and vitamin D supplements.

The therapeutic compliance valued by the MPR was 70% (Table 2), with no differences in the proportion of compliers, according to whether the treatment was weekly or monthly (68.2% vs. 73.5%, p=0.580). The compliance assessed by the self-administered Morinsky-Green questionnaire was 73%, with a moderate agreement compared to the MPR assessment (Kappa coefficient=0.436).

Table 2. Assessment of therapeutic compliance according to the MPR 

The mean value of the determination of CTX was 0.193±0.146 ng/ml (median=0.158 ng/ml), with lower patients compared to non-compliant patients (0.182±0.143 ng/ml vs. 0.2190±0.152 ng/ml; p=0.247). A cut-off point of CTX of 0.196 ng/ml was the one that presented a better sensitivity and specificity for the diagnosis of therapeutic compliance. Considering this CTX value, the therapeutic compliance was 64%, with a low concordance compared to the MPR assessment (Kappa coefficient=0.234). When considering the result of the Morinsky-Green questionnaire together with the value of the CTX, compliance was 51%, with a moderate agreement (Kappa coefficient=0.415).

Table 3 shows the values of sensitivity, specificity and predictive values of the different forms used to estimate therapeutic compliance. The area under the ROC curve (95% CI) for the Morisky-Green questionnaire was 0.7119 (0.6127-0.8111), and 0.6238 (0.5185-0.7291) for the evaluation by a CTX cut-off of 0.196 ng/ml (Figure 1). When considering the result of the Morisky-Green questionnaire together with the value of the CTX, the area under the ROC curve was 0.7452 (0.65730.8332), somewhat higher than if we only consider the Morisky- Green result (p=0.622) (Figure 2).

Table 3. Sensitivity values, specificity and predictive values for the different tools to estimate therapeutic compliance (95% CI) 

VPP: positive predictive value; NPV: negative predictive value; MG+CTX: Morinsky-Green and CTX.

Figure 1. ROC curves to assess therapeutic compliance using the Morisky-Green questionnaire (A), the CTX (B) and the Morisky-Green combination with the CTX (C) 

Figure 2. Comparison of ROC curves to assess compliance using the Morisky-Green questionnaire (MG) or the combination of MG and CTX 

DISCUSSION

In our sample of patients treated with the same bisphosphonate for at least the last year, a CTX determination of less than 0.196 ng/ml is an indicator of therapeutic compliance in the last year, with a moderate discriminating capacity, lower than the discriminative capacity of the Morisky-Green survey. Their joint assessment (CTX <0.196 ng/ml and Morisky-Green) improves the discriminative capacity, being a good option to assess the therapeutic compliance in the consultations. In a recent consensus document¹³, the initial and three-month determination of bone remodeling markers (CTX and P1NP) was recommended to assess non-compliance based on the observed change (decrease of 56% of CTX and 38% of P1NP). But this requires two determinations of two markers, which are not always accessible for primary care laboratories. In addition, it does not allow for assessing compliance in those patients who have already started treatment and there is no baseline available, nor does it allow analyzing noncompliance over time.

Their specific determination, along with the administration of a classic therapeutic compliance questionnaire, that of Morisky-Green, present the best sensitivity and the best negative predictive value for therapeutic compliance.

In our sample, the observed therapeutic compliance (measured according to the MPR) was high, 70%, much higher than that observed in our environment by different observational studies⁷. One of the possible explanations is that our study was not designed to assess the proportion of therapeutic compliance in our population and, therefore, random sampling was not carried out. In addition, more than half of the patients included had a previous fracture, although there were no significant differences in the percentage of patients with previous fractures between compliant and noncompliant subjects. The presence of previous fractures is associated with higher rates of therapeutic compliance¹¹.

One of every ten patients errors was observed in the correct way of taking the medication, a fact that in itself implies therapeutic noncompliance. These patients were not excluded ROC from the analysis since in all cases 1.0 the MPR was less than 0.8 and all 0.9 would be classified as non-compliant. As expected, the value of the CTX in these cases was not diminished since the absorption of the drug would be diminished. In the event that the MPR had been equal to greater than 0.8, the patient would have been considered as a non-compliant patient. As this situation has not occurred, 0.1 they have not been excluded from the study. Clear and concise information about the drug´s administration is required, as well as ensuring correct understanding of it, since an incorrect intake considerably decreases the absorption of the active principle and, therefore, the expected anti-fracture effect.

Unlike what was observed in previous studies carried out in the primary care field in Spain, where the diagnosis of osteoporosis was between 60-70%^17,18 and the densitometry before diagnosis was approximately 65%^17-19, in our study, both records were greater than 90%. This greater registry can be explained in part to a better registry of diseases and results over time, and to the fact that the patients included were assigned to doctors more aware of osteoporosis. This greater awareness of the professional with the condition could explain, in part, higher observed rates of compliance than that described in other population-based studies^7,1.

One of this study´s limitations is the way in which therapeutic compliance is valued through pharmacy billing data since they are not a direct indicator that the patient actually takes the medication, but exclusively that it withdraws from the medication. pharmacy. In the absence of direct methods to assess compliance, this is the most approximate and recommended measure to assess compliance.

Any patient with an MPR of less than 80% has been considered non-compliant, but not whether the non-withdrawal of medication occurred in the first or last months of the period prior to the CTX determination. This fact could have an impact in the CTX value.

Another limitation is that, once a CTX cut-off point is available to assess compliance, another patient sample should be checked to confirm that similar results are observed.

As a strength, the fact that it is a single CTX determination and that can be carried out at any time, together with the completion of a compliance questionnaire, facilitates better compliance assessment of patients treated with bisphosphonates, although they have been taking it for a long time.

CONCLUSIONS

The joint assessment of a single determination of the CTX and the Morisky-Green questionnaire presents a better discriminative capacity to assess therapeutic compliance. A CTX value of less than 0.196 ng/ml is the one with the best sensitivity and specificity.