The rate of prescribing gastrointestinal prophylaxis with either a proton pump inhibitor or an H 2-receptor antagonist in Nova Scotia seniors starting nonsteroidal anti-inflammatory drug therapy

1Faculty of Computer Science, Dalhousie University, Halifax, Nova Scotia; 2Division of Gastroenterology, University of Alberta, Edmonton, Alberta; 3Department of Medicine; 4College of Pharmacy, Dalhousie University, Halifax, Nova Scotia Correspondence: Dr Sander Veldhuyzen van Zanten, Division of Gastroenterology, Department of Medicine, 2-14A Zeidler Ledcor Centre, University of Alberta, Edmonton, Alberta T6G 2X8. Telephone 780-492-9840, fax 780-492-9865, e-mail vanzanten@ualberta.ca Received for publication November 18, 2009. Accepted December 8, 2009 origiNal arTicle

The rate of prescribing gastrointestinal prophylaxis with either a proton pump inhibitor or an H 2 -receptor antagonist in Nova Scotia seniors starting nonsteroidal anti-inflammatory drug therapy key Words: Cohort study; Cyclooxygenase-2 selective inhibitor; Drug utilization; Gastrointestinal prophylaxis; Histamine-2 receptor antagonist; Misoprostol; Nonsteroidal anti-inflammatory drugs; NSAIDs; Proton pump inhibitor; Prescribing; Seniors
T he risk of gastrointestinal (GI) complications attributable to nonsteroidal anti-inflammatory drugs (NSAIDs) varies according to the presence of one or more risk factors including advanced age, history of gastroduodenal ulcers, use of concomitant medications such as warfarin, acetylsalicylic acid and corticosteroids, concurrent diseases, use of higher doses of NSAIDs and duration of therapy (1)(2)(3)(4)(5).The RR of NSAID complications increases to 1.5 for individuals 60 to 69 years of age, and to 2.8 for patients 70 years of age and older compared with individuals younger than 65 years of age (6)(7)(8).
The recommended preventive treatment strategies for patients at increased risk of GI complications from NSAIDs are prophylaxis with gastroprotective agents (GPAs) or, alternatively, therapy with a cyclooxygenase-2 selective inhibitor (COXIB) with or without a proton pump inhibitor (PPI).Both approaches have demonstrated a similar risk reduction profile (9)(10)(11).The following GPAs have proven efficacy in decreasing the risk of gastroduodenal ulcers: once daily dosing with a PPI, misoprostol given in doses of 200 µg two to four times a day, and high-dose therapy with an H 2 -receptor antagonist (H 2 RA) (eg, famotidine 40 mg twice a day) (12,13).Importantly, the standard dose of H 2 RA has not been shown to provide adequate protection (14).However, NSAID prophylaxis is often not prescribed when it is indicated (15)(16)(17).Underprescribing of GPAs, especially in older patients or those with other risk factors, can lead to serious GI complications, mainly upper GI bleeding from ulcers, which can result in hospital admissions, the need for surgery and death.On the other hand, overprescribing GPAs and COXIBs for individuals without risk factors exposes patients to the side effects of these drugs (18,19) and unnecessarily increases health care costs.
The objectives of the present study were to describe the rate, timing and duration of GI prophylaxis in Nova Scotia seniors (individuals older than 65 years of age) receiving nonselective NSAIDs.

Study population
A retrospective drug claims database analysis of the Nova Scotia Seniors' Pharmacare Program (NSSPP) (www.gov.ns.ca/health/Pharmacare/seniors_pharmacare/seniors_pharmacare_q_n_a.asp) for the fiscal years 1998 to 2002 (April 1, 1998, to March 31, 2003) was conducted.More than 75% of Nova Scotia's seniors were eligible to participate in the NSSPP.During the period from 1998 to 2002, the number of eligible seniors in the NSSPP decreased from 92% (113,437 of 123,178) to 78% (101,010 of 128,908).This decrease was mainly due to a change in eligibility and cost-sharing requirements (20).
Beneficiaries pay a premium that is waived if the beneficiary receives the guaranteed income supplement from the federal government.Beneficiaries pay a copayment with a maximum annual copayment.All prescriptions are captured in the NSSPP database regardless of whether they are paid for by the Pharmacare program.

drugs studied
The WHO Anatomical Therapeutic Chemical codes and drug identification numbers were used to identify the drugs selected for the present study.All nonselective NSAIDs marketed in Canada are listed in the formulary of the NSSPP.All NSAIDs for which generic drugs were available were subject to the maximum allowable cost (MAC) policy.Under this policy, if the drug was available from multiple manufacturers, the maximum cost paid by Pharmacare was based on the lowest price among the interchangeable group of products.Beneficiaries must pay the difference between the total price for a drug and the MAC price for prescriptions covered under the MAC pricing policy.In addition, reimbursement of the MAC price started only when beneficiaries reached the yearly required deductible.Before reaching the deductible, beneficiaries must pay a portion of the entire cost of the prescription (16).
The COXIB drugs were included in the NSSPP formulary with a MAC stipulation on November 1, 1999, for celecoxib and on June 15, 2000 for rofecoxib.These two drugs were covered until rofecoxib was withdrawn from the market in September 2004.The use of acetylsalicylic acid was incompletely captured in this database because most seniors buy overthe-counter acetylsalicylic acid, which is less expensive than obtaining it by prescription.
Since 1992, PPIs have had specific criteria for reimbursement -in contrast to H 2 RAs and misoprostol, which have no restrictions (21).Specific criteria for NSAID prophylaxis with PPIs included the following: 1.For the treatment of NSAID-induced complicated peptic ulcers (bleeding ulcer, perforation, etc) when the NSAID is discontinued.Coverage duration: up to eight to 12 weeks.2. For the treatment and prophylaxis of NSAID-induced complications in patients who experienced previous NSAID-related ulcers or ulcer complications for which NSAID therapy cannot be discontinued.Coverage duration: while on NSAID or a maximum of one year with reassessment.3.For the prophylaxis of NSAID-induced complications in patients who are at high risk.
High risk was defined as receiving NSAID therapy plus two other risk factors including advanced age (older than 65 years), and concomitant anticoagulant or oral corticosteroid therapy.Coverage duration was for the duration of NSAID therapy up to 12 months, after which a reassessment was required.The age of the patient being older than 65 years, by itself, was not a sufficient criterion to qualify for coverage.

data analysis and patient cohort selection
The present study was approved by the Dalhousie Health Sciences Ethics Review Board (Halifax, Nova Scotia).Data were extracted from the Population Health Research Unit, which houses and administers the databases.Data were imported into a Microsoft Access (Microsoft Corporation, USA) database.Structured query language and visual basic were used to analyze the data and patient cohort selection.SAS (www.ansi.org/)(SAS Inc, USA) statistical software was used for data extraction and regression analysis.Several analyses were performed.

use rates of NSaIds and GPas
The use rates of NSAIDs and GPAs for all eligible Pharmacare beneficiaries were calculated.The yearly use rate of a particular drug was defined as the average of the monthly use ratescalculated by dividing the total number of individuals eligible for Pharmacare in the same month by the number of people using a particular medication.

Cohort selection Cohort of incident NSaId users (incident NSaId, prevalent GPa cohort):
A cohort of incident NSAID users was selected by eliminating all patients who never used an NSAID during the five-year study period.From the remaining sample, patients were required to be non-NSAID users in the 12 months before their index prescription (ie, the first NSAID prescription during the study period) and eligible for the Pharmacare program for at least 13 months (12 months before the index month plus at least the index month).
This cohort included patients who may have been taking a GPA at the time the NSAID was started.In this cohort, it was not possible to determine whether a decision was made to institute NSAID prophylaxis (ie, protecting the patients against upper GI complications) at the time the NSAID was started.

Cohort of incident NSaId and GPa users (incident NSaId, incident GPa cohort):
This cohort was created from the incident NSAID cohort by eliminating patients who used GPAs in the two months before their index NSAID prescription.It was assumed that a decision for NSAID prophylaxis was made at the time the GPA was dispensed or within the first month of the NSAID dispense event.By eliminating individuals who used GPAs in the preceding two months, confounding by patients who started the GPA for reasons other than NSAID prophylaxis (eg, because of gastroesophageal reflux disease symptoms), was avoided.In this cohort, the effect of NSAID therapy duration on the rate of coprescribing was also analyzed, as was the effect of increasing age.

duration of NSaId prescriptions
This analysis was performed in the incident NSAID incident GPA cohort.To classify NSAID prescriptions according to duration, it was necessary to eliminate prescriptions for which duration was uncertain.These prescriptions occurred either when patients were on an NSAID treatment course until the final month of the study period or when patients were on an NSAID treatment course when they lost their eligibility for the Pharmacare program.In both cases, these prescriptions may have continued but the data documenting their length were not available.After eliminating these 1560 prescriptions (original number of prescriptions = 18,265, with 16,725 after elimination), the proportion of NSAID prescriptions of a particular duration was calculated as the total number of prescriptions of a particular duration divided by the total number of treatment courses of all individuals in the cohort.An interval of at least one month without NSAIDs was used to separate the two treatment courses.The average duration of NSAID therapy was also calculated.Logistic regression was used to calculate whether the OR of being coprescribed a GPA during an NSAID treatment course increased with each month of duration of NSAID therapy and whether the patient's age at the start of the prescription influenced initiation of a GPA.

Coprescribing rates
In both cohorts, coprescribing rates were calculated according to the type of GPA (H 2 RA, PPI and misoprostol) and to age group.Coprescribing rates were calculated for each of the 48 months of NSAID prescriptions as the number of patients using one of the three GPAs in a particular month divided by the total number of NSAID users in the same month.
The prophylactic coprescribing rate is the rate at which GPAs were started at or within the first month (index month) as NSAIDs in the NSAID-GPA cohort.It is the coprescribing rate in the first month in the NSAID-GPA cohort.
A logistic regression analysis was performed to evaluate whether coprescribing changed with age.First, age was considered to be a continuous variable and calculated as the OR for being coprescribed in the index month over all age groups.All age groups were subsequently compared with the reference group (individuals 65 to 69 years of age).
In the present study, other risk factors for NSAID complications, such as the use of other antiplatelet agents, warfarin and corticosteroids, were not evaluated.

analysis of prescription claims
Pharmacare eligibility data and prescription claims were combined into person-month records of use.For each individual, their period of Pharmacare eligibility (or the study start and end dates if their eligibility was covered by the entire study period) was divided into 30-day increments (study month) and numbered from the start date of the study period (April 1, 1998).The start and end dates of each prescription -based on prescription filled date and the days supply field -were indexed to study months in the same manner.These two sets of person-month records were combined, with use flagged in the study months covered by a prescription.It was only necessary that a supply of drugs was available at any point during the month -it was not necessary for drug supply to be available for the entire month.
Individual patient prescriptions were aggregated into treatment courses based on prescription dates and the dispensed days supply of those prescriptions.Each prescription was assumed to start on the date it was filled, and ended on the fill date plus the days supply.For scenarios in which there was a gap between the end of the n th prescription and the start of the n th +1 prescription of less than 30 days, patients were assumed to be taking the NSAID and/or GPA over the entire period but at less than the prescribed daily dose.For instances in which there was a gap of greater than 30 days, it was assumed that the patient had stopped taking NSAIDs/GPAs and the treatment course had ended.A patient's treatment course was defined as the period from the initial prescription to the end of the last prescription with less than a 30-day gap.Within the treatment course, the available days supply was adjusted for 'hoarding' (ie, refilling a prescription before the expected end date) and 'stretching' (ie, refilling a prescription after the expected end date).For the final prescription in a treatment course, the expected end date was adjusted according to the average ratio of the dispensed days supply to the interval between the n th and n th +1 prescription observed within the treatment course.The start and end dates of each treatment course were used to define NSAID and/or GPA use within any study month.
Two methods were used to calculate the use rates for selected drugs.First, a yearly use rate was calculated by dividing the number of patients who had at least one drug prescription during the fiscal year by the total number of people eligible for Pharmacare in that year.Second, the yearly use rates were calculated as an average of monthly use rates, which were determined by dividing the number of people using a particular medication in a month by the total number of people eligible for Pharmacare in the same month.In the analysis, no adjustments were made for dose or duration of the NSAIDs prescribed.Any adjustments that may have occurred due to the changes in the eligibility criteria for seniors to receive Pharmacare coverage were not factored in the analysis.Information regarding the use of over-the-counter NSAIDs and H 2 RAs was not tracked.

duration of NSaId prescriptions
Only 2.3% of NSAID prescriptions had a continuous duration of more than one year, while 76% of prescriptions were for two months duration or less (Table 2).

Coprescribing rates of NSaIds and GPas
A total of 40,511 patients received at least one prescription for a nonselective NSAID during the five-year study period.Using the specified inclusion criteria, a cohort of 16,240 incident NSAID users (incident NSAID, prevalent GPA cohort) was selected, with 3334 (21%) of these patients already receiving a GPA at the time the NSAID prescription was started.After eliminating patients who used a GPA in the two months before the index NSAID prescription, the incident NSAID incident GPA cohort was formed with the remaining 12,906 incident NSAID incident GPA users.
Coprescribing rate according to GPa The monthly coprescribing rates of of NSAIDS and GPAs are shown in Figures 1 and 2. In the incident NSAID prevalent GPA cohort, the coprescribing rate for all GPAs in the first month was 21.2% of all NSAID users; 83.4% of all patients in this cohort were coprescribed an H 2 RA (Figure 1).In this cohort, coprescribing of H 2 RA increased from 17.7% of all NSAID users in the first month to 35.2% in month 48.PPI use increased from 3.7% of all NSAID users in the first month to 7.2% in month 39, and decreased to 4.2% in month 48.
In the incident NSAID incident GPA cohort, coprescribing rates were much lower.In the first month, the coprescribing rate for all GPAs (prophylactic coprescribing) was 3.8% of all nonselective NSAID users; 92.6% of all patients in this cohort were coprescribed an H 2 RA, 7% were coprescribed a PPI and 0.4% were coprescribed misoprostol.Logistic regression analysis showed that the OR of being coprescribed a GPA during an NSAID treatment course increased by 6% with each month of duration of NSAID use and by 10% with each five-year increase in age at the start of the treatment course.The rate of coprescribing with H 2 RAs increased with the duration of time patients were on NSAIDs, increasing from 3.5% of all NSAID users in the first month to 24.1% in month 48, while the rate of PPI coprescribing remained low, increasing from 0.3% to 1.9% of all NSAID users.The coprescribing of a PPI was very low and showed only a very small increase over the 48 months.Given that the average duration of NSAID prescriptions was short,  these data also indicate that GPAs were continued in many seniors even after the NSAID was discontinued.Therefore, it would be important to determine whether there was another reason for the continuation of PPI prescriptions.
Coprescribing rate according to age Table 3 shows the effect of age on GPA coprescribing in the incident NSAID prevalent GPA cohort, and incident NSAID incident GPA cohort.The rate of coprescribing in the index month increased with age but was less pronounced in the incident NSAID incident GPA cohort in which the prophylactic coprescribing rate was calculated.The OR of receiving NSAID prophylaxis was statistically significant compared with the reference group starting from age 80 years and older in the prevalent but not the incident GPA cohort.

dISCuSSIoN
A striking finding in our study was that, despite an overall high frequency of use of acid suppression in Nova Scotia seniors, the rate of gastroprophylaxis (ie, starting a GPA at the time the NSAID was initiated) with either an antisecretory agent (PPI, H 2 RA) or misoprostol was low during the study period (1998 to 2002).During the study period, no generic PPIs were available in the Canadian market, and many provinces -including Nova Scotia -required special authorization for use of PPIs in their seniors' Pharmacare programs.The fact that special authorization was required may be one explanation for the present study's findings; however, this could not be formally evaluated.Even taking this into account, the rate of coprescribing PPIs with NSAIDs seemed inappropriately low.Although reimbursement criteria for PPI use may be one explanation for the findings, there are other possibilities.One explanation may be a belief among Nova Scotia physicians that NSAID prophylaxis with GPAs is not required.This seems less likely because coprescribing rates in Quebec and other provinces were higher than in Nova Scotia.It is possible that physicians are insufficiently aware of the toxicity concerns of NSAIDs without GPAs and, hence, more education should be targeted to address this knowledge gap.Such education should also explain that prophylaxis with a standard-dose H 2 RA is insufficient.Computerized clinical decision support systems at the time of prescribing or dispensing the NSAID to patients at significant risk of GI complications may also be helpful.The NSSPP covers medication costs for approximately 80% of seniors living in Nova Scotia.
The use rate of antisecretory medication in this population is very high.The monthly rates of H 2 RA use varied from 13.6% to 15.2%, and PPI use increased from 1.8% to 4.6%.Overall, misoprostol use was very low (less than 1%).The reasons for this are unclear but are likely related to the more frequent need of dosing throughout the day (two to four times) and to the adverse effect profile, which, in clinical trials, led to a withdrawal rate of 8.5% (12).Despite the fact that overall GPA use was high, the striking finding in our study was that the rate of initiating a GPA at the time an NSAID was prescribed (NSAID prophylaxis) was low, although it did increase by 10% with each five-year increase in age at the start of the treatment course.The rate of coprescribing a GPA increased by 6% for each month that prescription of the NSAID was continued.The latter suggests that, at least to some extent, physicians more likely coprescribe a GPA with increased duration of NSAID use.We did not factor in any changes that may have occurred in the eligibility criteria for seniors to receive  Pharmacare coverage in our analysis.We also did not track information regarding changes in the use of over-the-counter NSAIDs.
It is well recognized that age is an important risk factor for NSAID complications.We did not analyze other risk factors for NSAID complications such as use of warfarin or corticosteroids, which may have influenced physicians' behaviours in deciding to prescribe GPAs as NSAID prophylaxis.In the United Kingdom (National Institute for Health and Clinical Excellence guidelines [22]) and the Netherlands (23), an age of 65 years or older is a sufficient criterion to receive NSAID prophylaxis; that is, providing the patient with either a PPI or misoprostol at the time the nonselective NSAID is started to decrease the risk of serious upper GI complications.Others (24) make similar recommendations including the American College of Rheumatology guidelines for the treatment of osteoarthritis (25).In Nova Scotia, age older than 65 years, in itself, is not a sufficient criterion to qualify for reimbursement of PPIs.In contrast, Nova Scotia has no specific eligibility criteria for the use of standard-dose H 2 RAs and misoprostol.This is possibly one explanation for the high use rate of H 2 RAs relative to PPIs.In our study, the use of NSAID prophylaxis (starting an antisecretory agent or misoprostol at the time that conventional NSAIDs are prescribed) was low -3.8% of all NSAID users.This rate was much lower than what has been reported in other studies (15,(26)(27)(28)(29) in which rates of coprescribing with gastroprophylatic drugs varied between 10.2% and 15% of NSAID users.The fact that in Quebec, where there is no restriction on PPI drug insurance coverage, the rate of coprescribing of a gastroprophylactic drug and an NSAID in a cohort of nonselective NSAID users older than 65 years of age was 10.2% suggests that the NSSPP-specific reimbursement criteria regarding PPIs may influence physician prescribing behaviour for patients receiving NSAIDs (28).An alternative explanation may be that with the easy availability of GPAs for NSAID prophylaxis in Quebec, physicians working there have developed a different attitude to this issue and, consequently, have a much lower threshold for prescribing NSAID prophylaxis.Of concern is also the fact that 92% of Nova Scotia patients received H 2 RAs while on NSAIDs.Of these H 2 RA prescriptions, more than 90% were for the standard dose (30).The literature contains convincing evidence that the standard dose of H 2 RA offers insufficient gastric protection against NSAID damage.In a randomized trial (14), the standard dose of ranitidine protected against the development of duodenal ulcers but not gastric ulcers.There was only one study (13) that used high-dose famotidine 40 mg twice a day, which demonstrated protection against both gastric and duodenal ulcers.However, at that dose, costs are higher than the once daily use of a PPI.Therefore, our results not only indicate that the rate of NSAID gastroprophylaxis in Nova Scotia was lower than that reported in other studies, but also that when gastroprophylaxis is initiated, the large majority are for the standard dose of H 2 RA, which provides insufficient protection.
Regarding age, not only is age older than 65 years clearly identified as a risk factor for NSAID complications, but advancing age increases the risk further (1)(2)(3)(4)6).In a multivariate analysis (7) of risk factors in the Misoprostol Ulcer Complication Outcomes Safety Assessment (MUCOSA) trial, age older than 65 years was an independent risk factor for serious GI bleeding events and the risk increased further with increasing age.Using age as a continuous variable in our study, we demonstrated a very small (but statistically significant) increase in the rate of prescribing with increasing age.However, considering five-year age strata in the incident NSAID incident GPA cohort, the increase in coprescribing was not significant even at age older than 85 years.This is in contrast to observations in the province of Quebec (28), where coprescribing rates clearly increased with advancing age.Our results demonstrated that the likelihood of patients receiving an antisecretory agent while they were taking NSAIDs increased as they aged.Whether this was because the patient became symptomatic with, for example, dyspepsia symptoms, which would trigger a prescription with an acid-suppressive agent, or whether the prescribing physician became more concerned with the increased duration of the NSAID is unclear.It would be interesting to know whether the low rate of NSAID prophylaxis in our population translates into an increased risk of serious complications such as bleeding ulcers, a need for surgery, hospitalization or death.This is a subject for further study.There are data from other studies that also indicate that NSAID prophylaxis is underprescribed in other populations (17,31).
Our data also showed that the duration of a substantial proportion of NSAID prescriptions were for a relatively short period of time, with 83% of patients using NSAIDS for three months or less.Only 2.3% of treatment courses lasted for more than 12 months.Whether this short duration of NSAID use is an explanation for the low NSAID prophylaxis rate is also unknown.When started in elderly patients, the duration of NSAID use is often unknown.Short duration should not influence the decision on whether to prescribe NSAID prophylaxis because the risk of complications starts early (2).A metaanalysis (2) showed that the risk of serious GI complications (hospitalization, bleeding, perforation and death) in NSAID users is greatest in the first month after the start of NSAIDs (OR=8 versus OR=1.9 after more than three months of NSAID use).
It is also well known that there is synergy between NSAIDs and acetylsalicylic acid use, and the increased risk of GI bleeding complications (5).Unfortunately, our database did not reliably capture acetylsalicylic acid use in this cohort because it is generally cheaper for patients to buy acetylsalicylic acid over the counter than to pay the pharmacist's dispensing fee and any applicable copayment fees.Because acetylsalicylic acid use in senior citizens is likely substantial, this would further increase the risk of GI complications and would be a further reason to use NSAID prophylaxis.
During the period from 1998 to 2002, the rates of nonselective NSAID use decreased while at the same time, the rates of COXIB use increased sharply after their introduction to the Canadian market in 1999.Importantly, there was an overall increase in the number of patients receiving either an NSAID or a COXIB.These trends for the Nova Scotia market are similar to those found in similar populations (age older than 66 years) in other jurisdictions such as Ontario (32).These data indicate that physicians are influenced by the type of medication that is available for a particular indication (eg, osteoarthritis).Whether this is due to effective marketing by pharmaceutical companies, restrictions on drug formularies, differences in side effects (eg, COXIB having a lower risk of GI bleeding complications) or other factors, is unknown.
Our study had several limitations.Patients excluded from the NSSPP due to private coverage may have limited representativeness, although over the five-year study period, coverage ranged from 78% to 92% of all seniors.As discussed, we could not evaluate the impact of the previous authorization policy on the rate of coprescribing.Our data cover the period from 1998 to 2002, and because generic PPIs are available and the authorization policy in Nova Scotia for PPIs has changed, the current rate of coprescribing may be higher.In addition, the awareness of the risk of adverse events related to NSAID use is likely much higher after all the publicity surrounding the use of COXIBs over the past few years.This may also have increased the awareness of the need for coprescribing.We did not assess the effect of switching from nonselective NSAIDs to selective COXIBs when calculating the use rates of all NSAIDs.Because COXIBs are associated with a lower rate of GI complications, this may have influenced NSAID prophylaxis coprescribing rates.We did not have access to other clinical diagnostic information and did not assess other risk factors such as use of warfarin or corticosteroids, which increase the risk of NSAID complications.Information regarding the presence of conditions such as gastroesophageal reflux disease, which can influence the use of GPAs, was also not collected.In Nova Scotia, some nonselective NSAIDs are available over the counter without a prescription, but this could not be assessed, nor could the concomitant use of acetylsalicylic acid, which may further increase the risk of serious GI side effects.The use of traditional nonselective NSAIDs was, therefore, probably underestimated in the present study population.We did not adjust for prescription channelling, in which patients with a different risk profile are more likely to end up on certain medications, which, in our study, would be PPI therapy.We did not investigate other risk factors such as use of other medications, anticoagulants or corticosteroids.We did not examine the cost-effectiveness of NSAID prophylaxis with a GPA.We also did not analyze whether any changes in the eligibility criteria for seniors to receive Pharmacare coverage occurred, nor did we study whether changes in eligibility influenced drug use; however, we believe it is unlikely that this would have significantly affected our results.Finally, our database reported dispensed medication and did not assess patient adherence to treatment.

Figure 1 )Figure 2 )
Figure 1) Coprescribing rate according to gastroprotective agent.Incident nonsteroidal anti-inflammatory drug (NSAID) cohort in Nova Scotia Seniors' Pharmacare Program benefiaries 1998 to 2002.*Calculated as a per cent of all NSAID users.H2RA H 2 -receptor antagonist; Miso Misoprostol; PPI Proton pump inhibitor The yearly use rates (based on average monthly use rates) of H 2 RAs, PPIs, misoprostol, NSAIDs and COXIBs for NSSPP beneficiaries are shown in Table 1.PPI use rates increased from 1.8% in 1998 to 4.6% in 2002.There was a slight increase in the use of H 2 RAs from 13.63% in 1998 to 15.21% in 2002.The use rate for misoprostol was very low (0.2% to 0.4%); factors affecting its use were not explored further.The rate of nonselective NSAID use decreased from 8.1% in 1998 to 5.0% in 2002.During this time, COXIB use increased from 0% in 1998 to a high of 6.4% in 2001.The total percentage of patients receiving either a nonselective NSAID or COXIB increased from 8.1% in 1998 to 11.5% in 2001.

TABLE 2 Duration of therapy with nonselective nonsteroidal anti-inflammatory drugs (NSAIDs) in Nova Scotia Seniors' Pharmacare Program beneficiaries (cohort of incident NSAID incident gastroprotective-agent users)
*Based on total number of treatment coursescohort of 12,906 incident nonselective NSAID incident gastroprotective-agent users followed for four years