Recent Trends in Sitagliptin Prescribing
Accurate unbiassed evaluation of the comparative safety and effectiveness of therapeutic agents in clinical practice requires a thorough evaluation of the characteristics of the patient populations being compared. When faced with a choice of agents having various levels of novelty, price, efficacy, safety and contraindications, clinicians rarely, if ever, prescribe them randomly. In general, many patient-specific factors (including demographic information and clinical history) will strongly influence the choice of agent prescribed. When similar sets of choices are made in large patient populations, confounding and bias can affect outcomes and need to be considered carefully in comparative analyses of observational data from clinical practice.
The analysis reported here is a follow-up to a previous evaluation of a large US insurance claims database which found that patients with T2DM who were using sitagliptin had higher proportions of comorbidities and greater use of prescription medications and physician visits compared with patients with T2DM using other OAHAs. The study window for the previous analysis was from 1 January 2006 to 30 June 2008, coincident with a period of time just after the approval of sitagliptin for marketing in the United States. To assess whether the pattern of preferential prescribing of sitagliptin to older patients with more comordities and diabetes-related complications has continued, we used the same database to evaluate characteristics of the patient populations receiving new prescriptions for OAHAs during a similar period (1 January 2006–31 December 2007) and during the 2 years and 3 months following (1 January 2008–30 April 2010).
The total number of patients with T2DM receiving a new prescription for OAHAs increased during the later time period studied compared with the earlier period, with the number of patients receiving new prescriptions for sitagliptin increasing disproportionately during this time (Table 1). This increase took place in every age group and in both genders to a similar extent (Table 1).
The overall increased number of new prescriptions for sitagliptin is reflected in the reduction in some of the differences in patients being prescribed sitagliptin compared with other subgroups. However, the differences in prescribing patterns observed during the earlier period examined persisted up to 4 years after approval for clinical use.
The oldest patient population evaluated (60–64 year olds) remained more likely to receive a new prescription for sitagliptin than younger patient groups, but the likelihood was reduced from 1.60- to 1.34-fold (Table 1), compared with other OAHAs. Patients already receiving one or more OAHA were also more likely to receive a new prescription for sitagliptin compared with patients on monotherapy (Table 1).
The most striking change in odds ratio between the two time periods studied was for patients with preexisting T2DM. During the later time period, patients with existing disease were slightly more likely to be given a new prescription for sitagliptin than patients with newly diagnosed disease, whereas during the earlier period they were less likely. In addition to increasing comfort of prescribing physicians with time, this change may reflect the availability of the sitagliptin-metformin fixed-dose combination in 2007, waning enthusiasm for prescribing thiazolidinediones (e.g. pioglitazone), or some combination of these factors.
Of the comorbidities and complications evaluated, patients with hypertension and proteinuria were more likely to receive a new prescription for sitagliptin than other OAHAs during both time periods evaluated. Although the ORs are generally consistent across the two time periods, patients with blindness, macular oedema or retinopathy had been more likely to receive a new prescription for sitagliptin than another OAHA during the earlier time period, but that difference was no longer apparent during the later time period. Patients with neuropathy became more likely to be given a new prescription for sitagliptin during the later time period, but that trend did not reach statistical significance during the earlier time period.
These observations suggest that trends in sitagliptin prescribing habits are changing, but that they remain somewhat biassed towards prescribing sitagliptin to older patients with more comorbidities and diabetes-related complications, consistent with our earlier report. This prescribing pattern is likely a reflection of the favourable efficacy and safety profile of sitagliptin, resulting in sitagliptin being more often prescribed to older patients as well as patients with concomitant medical conditions.
If not recognised and taken into account during observational evaluations of safety and effectiveness, the non-random prescribing pattern for sitagliptin can have significant consequences. Patients receiving sitagliptin, and possibly other DPP-4 inhibitors, may be at higher risk of clinical events simply because of increased risk as a result of baseline characteristics prior to initiation of therapy. At least one previous study highlighted the differences in prognosis associated with varying levels of comorbidities/complications in patients with type 2 diabetes. In that study, patients with higher levels of comorbidities received reduced cardiovascular benefit from intensive glucose control compared with those having lower levels. Thus, caution must be taken in the design, analysis and interpretation of comparative observational studies. The likelihood of channelling bias and confounding by indication must be evaluated, and acknowledged as a limitation of the study.
The database used for this study contains a large, unselected patient population and includes records of dates of healthcare encounters, and treatment and diagnostic codes for each patient. However, the database was not created for research purposes; it is primarily used as an administrative tool to facilitate payment and audit transactions related to healthcare provision. This can compromise some aspects of available data because diagnostic codes for some conditions may have been used to justify diagnostic testing to 'rule out' conditions; furthermore, diagnoses may be biassed towards those that provide higher reimbursement. Moreover, the types of conditions that can be studied are limited to those coded for billing and reimbursement purposes. In addition, the database is limited to people carrying commercial insurance and may not be representative of the entire treated population, such as Click here to enter text.elderly people of US population with Medicare insurance. In this study, a single occurrence of an ICD-9 code or a CPT code defined the presence of a baseline comorbidity. The codes have been carefully reviewed and defined by the physicians, but were not validated. However, for all subcohorts baseline comorbidities are likely to share the same limitations, therefore the analysis provides a good crude comparison of baseline comorbidities among patients with different OAHA treatments.
Discussion
Accurate unbiassed evaluation of the comparative safety and effectiveness of therapeutic agents in clinical practice requires a thorough evaluation of the characteristics of the patient populations being compared. When faced with a choice of agents having various levels of novelty, price, efficacy, safety and contraindications, clinicians rarely, if ever, prescribe them randomly. In general, many patient-specific factors (including demographic information and clinical history) will strongly influence the choice of agent prescribed. When similar sets of choices are made in large patient populations, confounding and bias can affect outcomes and need to be considered carefully in comparative analyses of observational data from clinical practice.
The analysis reported here is a follow-up to a previous evaluation of a large US insurance claims database which found that patients with T2DM who were using sitagliptin had higher proportions of comorbidities and greater use of prescription medications and physician visits compared with patients with T2DM using other OAHAs. The study window for the previous analysis was from 1 January 2006 to 30 June 2008, coincident with a period of time just after the approval of sitagliptin for marketing in the United States. To assess whether the pattern of preferential prescribing of sitagliptin to older patients with more comordities and diabetes-related complications has continued, we used the same database to evaluate characteristics of the patient populations receiving new prescriptions for OAHAs during a similar period (1 January 2006–31 December 2007) and during the 2 years and 3 months following (1 January 2008–30 April 2010).
The total number of patients with T2DM receiving a new prescription for OAHAs increased during the later time period studied compared with the earlier period, with the number of patients receiving new prescriptions for sitagliptin increasing disproportionately during this time (Table 1). This increase took place in every age group and in both genders to a similar extent (Table 1).
The overall increased number of new prescriptions for sitagliptin is reflected in the reduction in some of the differences in patients being prescribed sitagliptin compared with other subgroups. However, the differences in prescribing patterns observed during the earlier period examined persisted up to 4 years after approval for clinical use.
The oldest patient population evaluated (60–64 year olds) remained more likely to receive a new prescription for sitagliptin than younger patient groups, but the likelihood was reduced from 1.60- to 1.34-fold (Table 1), compared with other OAHAs. Patients already receiving one or more OAHA were also more likely to receive a new prescription for sitagliptin compared with patients on monotherapy (Table 1).
The most striking change in odds ratio between the two time periods studied was for patients with preexisting T2DM. During the later time period, patients with existing disease were slightly more likely to be given a new prescription for sitagliptin than patients with newly diagnosed disease, whereas during the earlier period they were less likely. In addition to increasing comfort of prescribing physicians with time, this change may reflect the availability of the sitagliptin-metformin fixed-dose combination in 2007, waning enthusiasm for prescribing thiazolidinediones (e.g. pioglitazone), or some combination of these factors.
Of the comorbidities and complications evaluated, patients with hypertension and proteinuria were more likely to receive a new prescription for sitagliptin than other OAHAs during both time periods evaluated. Although the ORs are generally consistent across the two time periods, patients with blindness, macular oedema or retinopathy had been more likely to receive a new prescription for sitagliptin than another OAHA during the earlier time period, but that difference was no longer apparent during the later time period. Patients with neuropathy became more likely to be given a new prescription for sitagliptin during the later time period, but that trend did not reach statistical significance during the earlier time period.
These observations suggest that trends in sitagliptin prescribing habits are changing, but that they remain somewhat biassed towards prescribing sitagliptin to older patients with more comorbidities and diabetes-related complications, consistent with our earlier report. This prescribing pattern is likely a reflection of the favourable efficacy and safety profile of sitagliptin, resulting in sitagliptin being more often prescribed to older patients as well as patients with concomitant medical conditions.
If not recognised and taken into account during observational evaluations of safety and effectiveness, the non-random prescribing pattern for sitagliptin can have significant consequences. Patients receiving sitagliptin, and possibly other DPP-4 inhibitors, may be at higher risk of clinical events simply because of increased risk as a result of baseline characteristics prior to initiation of therapy. At least one previous study highlighted the differences in prognosis associated with varying levels of comorbidities/complications in patients with type 2 diabetes. In that study, patients with higher levels of comorbidities received reduced cardiovascular benefit from intensive glucose control compared with those having lower levels. Thus, caution must be taken in the design, analysis and interpretation of comparative observational studies. The likelihood of channelling bias and confounding by indication must be evaluated, and acknowledged as a limitation of the study.
The database used for this study contains a large, unselected patient population and includes records of dates of healthcare encounters, and treatment and diagnostic codes for each patient. However, the database was not created for research purposes; it is primarily used as an administrative tool to facilitate payment and audit transactions related to healthcare provision. This can compromise some aspects of available data because diagnostic codes for some conditions may have been used to justify diagnostic testing to 'rule out' conditions; furthermore, diagnoses may be biassed towards those that provide higher reimbursement. Moreover, the types of conditions that can be studied are limited to those coded for billing and reimbursement purposes. In addition, the database is limited to people carrying commercial insurance and may not be representative of the entire treated population, such as Click here to enter text.elderly people of US population with Medicare insurance. In this study, a single occurrence of an ICD-9 code or a CPT code defined the presence of a baseline comorbidity. The codes have been carefully reviewed and defined by the physicians, but were not validated. However, for all subcohorts baseline comorbidities are likely to share the same limitations, therefore the analysis provides a good crude comparison of baseline comorbidities among patients with different OAHA treatments.
Source...