The safety of levothyroxine use with respect to bone mineral density

Background: Previous studies on bone mineral density (BMD) abnormalities associated with hypothyroidism are scarce and not conclusive. The effect of thyroid hormone therapy on BMD has shown mixed results. 
Aim: This study aimed primarily to determine the association between the positive history of levothyroxine administering and the risk of osteoporotic fracture in Jordanian cohort of both genders, including post-menopausal women, who were attended to our rehabilitation clinic.
Methods: This study trial was an observational study, which was conducted retrospectively at Prince Rashid bin Al-Hasan Military Hospital, Royal Medical Services, Irbid, Jordan. The Binary Logistic Regression (BLgR) analysis was conducted for the 2 contrarily Levothyroxine (Tx) comparative group; Tx dependent cohort (Cohort II) versus Non-Tx dependent cohort (Cohort I), against the probability of being on the higher (versus lower) risk of femoral hip osteoporotic fracture (fHOPF). The studied patients were dichotomously categorized into 2 comparative cohorts; non-Levothyroxine dependent cohort [Cohort I] versus Levothyroxine dependent cohort [Cohort II]. A Chi Square test was processed across these 2 dichotomized cohorts to express the comparison results as Number (Percentages), strength of associations (odd ratios), Pearson chi-square statistic (χ 2), Goodness of Fit (G-Test of independence), and Pearson (r) and Spearman (ρ) correlations.
Results: The BLgR analysis results revealed that the unadjusted risk ratio for the higher probability of fHOPF in our investigated patients, who were on Thyroxine therapy [Tx dependent cohort, Cohort II] compared to the Tx independent cohort [Cohort I], was 10.969 (95% CI; 4.615-26.072). The explained variations in the fHOPF risk related Tx dependent status on our model of this BLgR based model ranged from 17.9%-23.8% (depending on the inferential Cox & Snell R2 or Nagelkerke R2 methods, respectively) and correctly classified approximately 61.2% of the overall cases. The constructed BLgR model was formulated as {e (-0.627+2.395×Tx)/[1+ e (-0.627+2.395×Tx)]}.
Conclusion: We revealed that the patients who were used thyroxine therapies (Cohort II patients) had significantly higher proportional distribution of higher risk of fHOPF compared to Cohort I patients [41 (85.4%) vs 55 (34.8%), respectively] with an odd ratio of 10.969 (95% CI; 4.62-26.07) and significant positive correlation of 0.429±0.057, χ2=37.889, p-value=0.000.