In this large population-based study, a significant drop in PTX rates among patients receiving dialysis after January 1, 2006 is shown. This drop coincides with cinacalcet introduction, but an increasing trend in the use of phosphate binders (but less calcium-based binders) and vitamin D across the time period is also shown.
Our results are consistent with a pooled analysis of short duration trials that showed a reduce incidence of PTX with cinacalcet (from 41 /1000p-y in the placebo group to 3 /1000 p-y in the cinacalcet group)
. More recently, a large placebo-controlled trial designed to evaluate cardiovascular endpoints showed a 69% decreased risk of PTX in the cinacalcet group
. Few studies have evaluated trends in PTX rates, and only one has reported rates after the release of the KDOQI clinical practice guidelines in 2003, and none have included patients well beyond introduction of calcimetics
[5–8]. Li et al.
 described PTX rates in the US from 1992 to 2007. A PTX peak rate was found in 2002 (12.8 /1000 py) followed by a decreasing rate until 2005 (5.4 /1000 py) and then an increase again in 2006–2007 (8.6 /1000 py)
. The current data shows a different pattern where the peak annual PTX rate was in 2004 (16.0 /1000 py) instead of 2002, followed by a decrease in 2006 (3.9 /1000 py). After 2006, annual PTX rates remained between 3.0 and 4.6 /1000 py. While January 2006 was the only statistically significant cut-off point in our model, visually the change in rate appeared slightly before that and all cut-off dates between January 2005 and June 2006 showed a trend for a lower PTX rates. This is consistent with the fact that cinacalcet market penetration was progressive over that period of time, and the lack of statistical power may explain why most were not statistically significant. Despite not being widely available before June 2006, carefully selected patients could have received cinacalcet from Amgen or private insurance companies. In the context where cinacalcet was probably given to patients who would otherwise have undergone a PTX and because the annual counts of PTX ranged from 29 to 57 before 2005, only a few patients receiving cinacalcet could have highly influenced the observed PTX rates. For this reason, it is not unexpected to see a flat trend for PTX rates after 2006 while cinacalcet market share continued to increase constantly. Because severe secondary hyperparathyroidism may not be completely controlled in all patients using cinacalcet and because some patients may not tolerate cinacalcet or would prefer to undergo a PTX, it is expected that rates of PTX would not fall indefinitely. It is unknown why PTX rates did not increase in 2006–2007 as was observed in the US. However, it is speculated that wider availability of cinacalcet through reimbursement in Québec may be a factor. It is also possible that, despite the KDOQI guidelines opinion-based recommendation for PTX in patients “with severe hyperparathyroidism (persistent serum levels of intact PTH >88.0 pmol/L [800 pg/mL]), associated with hypercalcemia and/or hyperphosphatemia that are refractory to medical therapy”
, Canadian physicians may be more inclined to wait for symptomatic hyperparathyroidism before sending a patient for a PTX. The higher PTX rates observed here among prevalent and younger patients are consistent with previous studies and with the physiopathology of secondary hyperparathyroidism
The portrait of phosphate binders and vitamin D use has changed in the last decade. This data shows a clear shift to lower dosing of calcium-based binders and an increasing utilization of non-calcium-based binders such as sevelamer and lanthanum. While the average dose of vitamin D remains fairly low (alfacalcidol increased slightly and calcitriol remained stable), the proportion of prescribed vitamin D increased considerably, reaching 70% in 2010. Unfortunately, only intravenous vitamin D analogues were available in Li et al.
 study, therefore use of phosphate binders, cinacalcet and oral vitamin D was not reported. Nevertheless, this switch from calcium-based to non-calcium based phosphate binders is expected after the publication of studies associating high calcium-phosphorus product with a higher risk of mortality and cardiovascular calcifications
[21, 22]. In 2003, KDOQI guidelines
 recommended a maximum dose of 1500 mg/day of elemental calcium, which was significantly lower than the usual practice. The increased of phosphate binders and vitamin D within this RAMQ data is also probably explained by the impact of KDOQI Guidelines on clinical practice and targets for serum phosphorus and PTH, and initial vitamin D dose. Additionally, cinalcalcet reimbursement was conditional on specific criteria as detailed previously, and physicians were encouraged to use vitamin D and phosphate binders to gain access to cinacalcet for their patients.
This study reports on changes of PTX trends over time, but cannot establish a causal link between cinacalcet and lower PTX rates. A stronger design would have been to compare PTX rates in patients receiving cinacalcet with rates in patients who did not. However, in the absence of available PTH levels, it was not possible to identify a control group that would have had the same severity of hyperparathyroidism and did not receive cinacalcet. Therefore, results would have suffered from an indication bias. Change in rates may be due to other factors occurring simultaneously with the introduction of cinacalcet. For example, following the publication of the KDOQI Guidelines use of more aggressive treatments such as higher doses of phosphate binders and vitamin D may explain lower PTX rates with time
[1, 19]. Decreased use of calcium-based phosphate binders may also explain decreased PTX rates by decreasing serum calcium, which would allow increase vitamin D use. Another explanation would be a change of indication or physician attitude toward PTX. For example, physicians may now tolerate higher PTH level thresholds before considering a PTX. However, these clinical practice changes are usually not acute in time and cinacalcet introduction remains the most probable explanation for the reduction in PTX rates observed. Despite the fact that the drop in PTX rates is visually convincing during the intervention period, we tested six different intervention dates and statistically significant results may be due to chance by multiple testing (limited power precluded applying correction factors such as Bonferroni). The unadjusted PTX rates are presented, since small numbers precluded standardization. An aging population may lead to lower PTX rates while an increasing proportion of prevalent patients may increase PTX rates. However, the population characteristics did not change much during the study years and the same PTX trend pattern was seen among subgroups (Figure
3). Moreover, adjusting PTX rates in the Li et al.
 study did not make a significant difference in the time period of the current study.
Some PTX procedures may have been missed if they were not coded. However, claims are required from surgeon compensation and it would be unlikely that a PTX would have been missed. RAMQ codes for PTX did not change during the study period. If any percutaneous ethanol ablation procedures of the parathyroid were carried out, these would not be captured, as there is no specific code for this new technique. But this technique was carried out on one or two anecdotal cases only and cannot explain the reduced rates. As with most retrospective studies using dispensed medications, there is uncertainty with patient compliance. In addition, intravenous vitamin D analogue claims are not captured within the RAMQ database, since vitamin D analogues are dispensed by the dialysis unit and a claim would not be submitted by the patient to the RAMQ. However, intravenous vitamin D is not commonly used in Québec, in part for economic reasons and health care organization. Finally, medication use among patients covered by private drug plans who are generally younger patients (an important risk factor for severe secondary hyperparathyroidism) could not be assessed. However, the main analysis on PTX trends included all patients regardless of age and availability of private insurance. Moreover, because the ARIMA model relied on time points and not on patient-level exposure, incomplete ascertainment of exposure does not influence our results. Finally, this study is limited to a single Canadian province, and results may not be generalized to other populations with different healthcare practices.