Doctor pharmaceutical utilization behaviour changed by the global budget programme strategies on hypertensive outpatient prescription

Doctor pharmaceutical utilization behaviour changed by theglobal budget programme strategies on hypertensiveoutpatient prescriptionjep_1550 262..268

Ching-Kuo Wei MPH,1 Shun-Mu Wang MHA1 and Ming-Kung Yeh PhD2

1Assistant Professor, Department of Health Care Administration, Oriental Institute of Technology, Pan-Chiao City, Taipei County, Taiwan2Professor, Department of Pharmacy and Institute of Preventive Medicine, National Defence Medical Center, Taipei City, Taiwan

Keywords

drug utilization evaluation, global budget,hypertension, national health insurance

Correspondence

Prof Ming-Kung Yeh161, Sec. 6MingChen E. RoadTaipei City 11490TaiwanE-mail: [email protected]

Accepted for publication: 22 July 2010

doi:10.1111/j.1365-2753.2010.01550.x

AbstractRationale, aims and objectives This study was to examine changes in doctor pharmaceu-tical utilization behaviour in response to Taiwan’s newly implemented National HealthInsurance individual hospital global budget (GB) programme and the changes in health carecosts and prescription trends for hypertensive (HT) patients.Method We analysed hospital outpatient prescription utilization with a pre–post individualhospital GB group and comparison group (the hospitals who did not join the programme)to evaluate the impact of GB strategies on hypertensive expenditure. Descriptive analyseswere performed based on the average daily medication expenditure for each prescription,and average number of items per prescription.Results This study reviewed 16 770 057 outpatient records and prescription records of213 568 hypertensive patients. The average total medication expense (+17.6%), HT medi-cation expense (+8.8%), daily medication expense (+16.3%), and daily HT medicationexpense (+6.3%) significantly increased after the action. After the individual hospital GBaction, hospital doctors participating in action switched their patients’ prescription drugs toother less expensive drugs such as rennin-angiotensin-aldosterone system inhibitors(-1.1%). The increase in volume of medications prescribed for control group were signifi-cantly larger for both alfa- and beta-adrenergic blocking agents (1.5%), and calciumchannel blocking agents (3.9%).Conclusion The individual hospital GB programme slowed down the trend of prescriptiondrug cost increasing rate and reduced the prescription drug volume in hospitals.

IntroductionTaiwan’s National Health Insurance (NHI) Scheme is the majorsource of national health financing, responsible for 97% of thepopulation [1]. All hospitals in Taiwan offer outpatient care andhave on-site pharmacies. Patients visiting hospitals for outpatientcare obtain their prescription drugs from these hospital pharma-cies. Most clinics also have on-site pharmacies, and doctors inthese clinics prescribe and dispense drugs [1]. Prior to 2002, insur-ers paid hospital providers on a fee-for-service basis, on a fixed-feeschedule, with patients free to choose between providers. The highproportion of NHI spending on pharmaceuticals has always beenregarded as a problem in the system. According to NHI estimates,the proportion of NHI expenditure on pharmaceuticals was around25% each year from 1996 to 2002 and about 28% in 2003 [2–5].Pharmaceutical expenditure (PE) of the NHI programme inTaiwan increased from 62.2 billion New Taiwan Dollars (NT$) in

1996 to NT$ 94.5 billion in 2003 [2–5]. The government hasintroduced several strategies to control PE since the inception ofNHI, including price adjustments based on the prices of interna-tional products or existing products (inter-brands comparison), ormarket price and volume survey; delegation of financial responsi-bility to regional bureaus; co-payment for outpatient drugs;generic grouping (the reference pricing scheme based on chemicalequivalence); a global budget (GB) payment system for clinics andhospitals; and reduction in the clinics’ daily payment rate fordrugs.

Factors affecting prescribing relate to the patient and society,medication, prescriber, practice environment and organization,available information and other external factors. No singleapproach is appropriate for every prescribing problem, health pro-fessional prescriber practice or health care setting. However,reports show that certain factors affect prescribing, includingstories in the media; drug reimbursement policies of government

Journal of Evaluation in Clinical Practice ISSN 1365-2753

© 2010 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice 18 (2012) 262–268262

and private drug plans and the associated workload for prescribers;government policies on doctor remuneration; standards of practicefrom professional organizations; prescribers’ concerns about legalliability, regulatory and control measures; and political consider-ations [6,7]. In order to promote safe, effective and efficient druguse, it is important to recognize the many interacting factors thatinfluence decision making in the medication-use system. A 2005Cochrane Review [8] listed interventions tailored to change pro-fessional health care practice classification of barriers, includinginformation management, clinical uncertainty, sense of compe-tence, perceptions of liability, patient expectations, standards ofpractice, financial disincentives, administrative constraints andothers.

This study investigates doctor pharmaceutical utilization behav-iour changed by the effects of Taiwan’s reimbursement rate adjust-ment based on individual hospital GB in July 2003. The presentstudy takes advantage of a rare policy analysis opportunity, com-paring policies of the pre- and post-individual hospital GB pro-gramme in Taiwan and examines the effects on outpatientmedication usage. This research also demonstrates the use ofmicro-level data to generate policy-relevant information that canbe used to improve efficiency in the use of health care resources.

Methods

Study objectives, study periods and population

The Taiwan individual hospital GB programme was launched inJanuary 2003. This study aimed to investigate changes in the dailyexpenditure and changes in prescribing from a year prior to thehospital GB action (the pre-action period is January 1, 2002–December 31, 2002) to a corresponding 1-year period after theaction (the post-action period is January 1, 2004–December 31,

2004) in a hospital setting. Figure 1 is the study flow chart. Themajor policy difference between the two study periods was theimplementation of the individual hospital GB action. Therefore,this study does not have disentangling effects from policies con-currently applied to control drug expenditures.

Our samples consisted of all patients from 26 hospitals, includ-ing five medical centres and 21 general hospitals in the Taipei area,the largest metropolitan area in the country with a population ofover 5 million. The patients were each receiving clinical services atthe same hospital during the years 2002 and 2004, and docu-mented patients were classified as indicated by the presence ofhypertensive (HT) and ICD-9-CM codes (401–405), respectively.

Study design and hypotheses

Patients were assigned to either an exposure group or a compari-son group. When a hospital joined the individual hospital GBprogramme, patients were included in the exposure group. If theirdoctors switched their patients’ prescription drugs to other, lessexpensive drugs after the action or fewer items per prescription, wewould expect to observe a reduction in their daily anti-hypertensive drug expense, volume of expensive anti-hypertensivedrugs used and average number of items on each prescription. Weanticipated that the extent to which doctors would switch theirpatients’ drugs to other more expensive drugs would be limited,and thus hypothesized that we would observe a reduction in the useof more expensive anti-hypertensive drugs, and also reduced dailyanti-hypertensive drug expense for the exposure group. Patientsbeing treated in hospitals that did not join the individual hospitalGB programme were included in the comparison group. Withcomparison group patients, the doctors had no obvious reasons toswitch the prescription drugs to less expensive drugs. We thushypothesized that we would observe an upward trend in the daily

Study period (Pre-action) Study period (Post-action)

2002ceD2002naJ

Individual hospital GB implement July 2003

4002ceD4002naJ

July 2003 Individual hospital GB

implement

Control group Exposure group

Definition of dependent and independent

variables

Data analysis

Definition of dependent and independent

variables

Data analysis

Not-join hospital Join hospital

Population select: 1. HP ICD-9-CM codes :401–405 2. During 2002–2004, the patient

were visited the same doctorat the same hospital.

3. The medication expenditure > 0 4. The duration of the prescription

> 7 days 5. The gender data was available 6. The age within 0–100

Figure 1 Study flow chart.

C.-K. Wei et al. Doctor’s behaviour changed by the GB on HP

© 2010 Blackwell Publishing Ltd 263

drug expenditure for this patient group. Regarding consumptionvolume, we hypothesized that doctors would change anti-hypertensive medicines from more expensive to less expensivesuch as from rennin-angiotensin-aldosterone system inhibitors(RAASI) to alfa- and beta-adrenergic blocking agents (ABAs) orcalcium channel blocking agents (CCBs). This change in prescrib-ing would decrease the volume of RAASI drug prescriptions forthe exposure group, but increase the volume of RAASI drug pre-scriptions for the comparison group, since the NHI drug reim-bursement scheme operates according to a fee-for-services model.We also hypothesized that the total drug expenditure wouldincrease for both patient groups.

Data sources and analysis

The outpatient records were first entered into a data bank with thestring filing system, as provided by the Bureau of National HealthInsurance and ethics committee, Taipei Branch (NO. 0940002727)(about one-fourth of Taiwan population is enrolled) between years2002 and 2004. The following files were included: outpatientprescription and treatment details, outpatient prescription anddoctor order details, contracted pharmacy prescription dispensingdetails and contracted pharmacy prescription dispensing doctororders. Prescriptions for anti-hypertensives were selected when thefollowing criteria were met: (1) medication expenditure claim was>0; (2) duration of the prescription was >7 days; (3) gender data ofthe patients were available; (4) age of patients was between 0 and110 years; and (5) patients were being medicated at a singlehospital in Taipei Branch during the study periods. Each recordcontained information about basic characteristics of the hospitaland the patient, the date of the visit, the major ICD 9 code (Inter-national Classification of Diseases, Ninth Revision) for this visit,and the prescription duration. Each record also contained informa-tion on the name of the prescription drug, and the pharmacyexpenditure. Based on these outpatient visit records, we furtherconstructed a data file comprising patient-hospital-period records.Each record pertained to a patient’s use of pharmaceuticalsobtained from a specific hospital during the pre-action or thepost-action period.

The data were obtained from the files kept for every Taiwanesecitizen using an assigned identification number. The data includeditems including the identification number of documented patients,age, sex, number of visits, frequency of admission and fees fordifferent health care activities. Detailed information about thecoding systems and further information of the hospital character-istics were obtained from the health research institute. We linkedthis data to corresponding registration data of health care facilitiesin Taiwan and an archive constructed by the NHI to record allhistorical data on the reimbursement rates of drugs covered by theNHI.

Throughout this study, our reference to prescription cost refersto ‘price NT$’ of the government data in July 2004. Prescriptioncost does not refer to cost from the manufacturer or seller’s per-spective, that is, not the amount that is paid to the manufacturer orseller for the drug. In addition, because of the descriptive nature ofour analysis, we focused on changes in drug utilization followingthe policy shift instead of on comprehensive policy effects. Thedefinition of a prescription requires special clarification. In theUSA, a prescription is a doctor’s order for a single drug. Multiple

medications for one condition may result in prescriptions for eachdrug. For example, if a doctor prescribes four drugs for a patient,there are four prescriptions. By contrast, a Taiwan prescriptionusually contains orders for multiple drugs. The numbers of drugsin a prescription are referred to as drug items. In other words, theabove-mentioned US example would be considered one prescrip-tion with four drug items in Taiwan. Throughout the paper ourreference to a prescription is based upon the Taiwan definition.

Dependent and independent variables

We focused on pre-selected drug utilization measures for indi-vidual matched patient profiling of outpatients. The definitions ofthe five outcome variables for this study are as follows: (1) Totalexp – the total drug expenditure corresponding to a specificpatient-hospital-period record; (2) Total exp(HP) – the total anti-hypertensive drug expenditure corresponding to a specific patient-hospital-period record; (3) Day exp – drug expenditure per daycorresponding to a specific patient-hospital-period record; (4) Dayexp(HP) – anti-hypertensive drug expenditure per day correspond-ing to a specific patient-hospital-period record; and (5) anti-hypertension drug (HP) items: average volume of items for eachprescription containing HP. The volume of drug items used wascounted according to the Anatomical Therapeutic Chemical (ATC)classification system with C02, C03, C07, C08 and C09 for hyper-tension. For each ATC class of drugs, we then constructed a datafile consisting of records of outpatient visits to hospitals for treat-ing hypertension for each patient group described above.

Units of analysis and statistical analysis

We constructed a longitudinal database for this study. Eachpatient–hospital group in the database was treated as a cluster andcontained two records: one for the pre-action period and the otherfor the post-action period. This allowed us to apply panel dataestimation methods to estimate the influences of the action. Thestatistical package for the social sciences (spss/Pct, version 9.0)was used for data entry and analysis.

ResultsThe study focused on analysing the details of outpatient prescrip-tions. The pharmaceutical expenditures for anti-hypertensivemedicines during NHI individual hospital GB programme implan-tation showed that the monthly costs of prescribed anti-hypertensive drugs accounted for 65.4% of the total amountapproved for medications. Anti-hypertensive monotherapy wasbeing taken by 43.0% of hypertensive patients. The drugs mostfrequently prescribedas monotherapy were angiotensin convertingenzyme (ACE) inhibitors (39.7%), calcium channel blockers(25.4%), beta-blockers (19.6%), diuretics (10.6%), and others(4.7%). For combination therapy, ACE inhibitors and diuretics(32.6%) were most frequently prescribed and used. Overall, themost used anti-hypertensive drugs were ACE inhibitors. Resultsshowed that the average daily medication expenditure per pre-scription for the HT group fluctuated slightly, but was stable overtime. The categories of average daily medication expenditure andthe average daily anti-hypertensive medication expenditure bothdisplayed increasing trends.

Doctor’s behaviour changed by the GB on HP C.-K. Wei et al.

© 2010 Blackwell Publishing Ltd264

Figure 2 shows the distribution of HT outpatient prescriptionsacross gender and age groups in our sample. The 16 770 057 HToutpatient records of the study sample are included; 8 049 984records of prescriptions were from the same single hospital, and5 937 581 records (213 568 patients) were from hospitals that hadjoined the individual hospital GB during the 3 years and these areultimately selected for use in the final analysis. The sampleincludes 108 142 males with an average age of 63.5 � 13.5 yearsand 105 426 females with an average age of 64.5 � 12.2 years.

Table 1 summarizes the means of the outcome variables forpatient groups over the two periods. The point estimates of themeans show that the total medication expense, HP medicationexpense, daily medication expense, and daily HP medicationexpense significantly increased after the action, but the averagevolume of prescribed drug items decreased. As hypothesized, totalexpenses increased significantly (17.6%) from NT$1226 toNT$1442 after the action. The individual hospital GB programmeimplemented by the NHI in Taiwan did not reverse the trend ofincreasing prescription drug costs in hospitals. Comparing thepost-GB group with the pre-GB group, the average daily medica-tion expenditure increased 16.3% from NT$49 to NT$57, and theaverage daily anti-hypertension medication expenditure onlygradually increased by 6.3% from NT$32 to NT$34. The numberof items contained in each prescription decreased by 5% from 2.1to 2.0.

Table 2 shows the changing pattern in numbers and types ofdrugs used between the pre-action period and the post-actionperiod based on ATC classification. During the pre-action period,the three most prescribed drugs in order of preference were ABAs,RAASI, and CCBs, but during the post-action period the mostprescribed drugs changed to ABAs, CCBs and RAASI, respec-tively. The average extent of decreases in prescribed drugs, shownin order of volume, were diuretics, RAASI, antilipemic drugs,other drugs, cardiac drugs and vasodilators. In contrast, thenumber of CCBs, ABAs, and antithrombotic agents increased to402 270, 111 353, and 118 789, respectively. Doctors switchedtheir prescription drugs to other, less expensive drugs (ABAs andantithrombotic agents) after the action. However, comparingRAASI drugs prescribed for the pre-action period with those in thepost-action period, prescriptions for these significantly decreasedby 1.1% (177 083 total number prescribed). As hypothesized, theswitching of patients’ drugs by doctors to other more expensivedrugs was limited. In the post-action results, the reduction rate fordaily HP drug expense was greater than the rate for total dailydrugexpense, and reductions in average number of drug items on eachprescription.

The volume of drug use significantly decreased from the pre-action period to the post-action period for the exposure group(Table 3). Based on ATC classification, the volume of diuretics,vasodilating, antilipidemic, and cardiac drugs used decreased, and

Figure 2 Demographic distribution of hyper-tensive patients in north Taiwan (2002–2004).BP_M, male; BP_F, Female.

Table 1 The outcome of the change inexpenditure and items after the action

Pre-action Post-actionChange afterthe action

MeanStandarddeviation Mean

Standarddeviation Difference %

Total exp (NT) 1226 1135 1442 1221 216 17.6Total exp(HP) (NT) 853 726 928 750 75 8.8Day exp (NT) 49 52 57 52 8 16.3Day exp(HP) (NT) 32 25 34 26 2 6.3HP items 2.1 1.3 2.0 1.1 -0.1 -5.0

The global budget hospitals of pre- and post-action outcome for total drug expenditure (Total exp),total anti-hypertensive drug expenditure [Total exp (HP)], drug expenditure per day (Day exp),anti-hypertensive drug expenditure per day [Day exp (HP)], and average volume of items for eachprescription containing anti-hypertension medicines (HP items).HP, anti-hypertension drug; NT, New Taiwan Dollar.



the prevention or indirect anti-hypertensive drug expense signifi-cantly decreased from the pre-action period to the post-actionperiod for both groups. In contrast, for the comparison group, thevolume of drugs used was significantly increased for both ABAsand CCBs. For the exposure group, the volume of CCB and ABAdrugs used increased by 3.3% and 2.1%, respectively. For thecomparison group, the volume of drugs used for these two classeswere 4.1% and 2.0%, respectively. However, for the exposuregroup, RAASI is a more expensive anti-hypertensive medicine,

and the volume used was significantly decreased by 1.0% (32 332)from pre-action to post-action. In the comparison group, no sub-stantial difference was found (0.0%), but a slight increase involume (9692) was noted. For the CCB class of drugs, the volumeof drug used was significantly increased for the comparison group.As hypothesized, hospital doctors in hospitals that joined the indi-vidual hospital GB programme switched their patients’ drugsto other less expensive drugs such as from RAASI to CCBs(Table 3).

Table 2 The difference in the medicine usednumber between two periods based on ATCclassification

Agents


Volume % Volume % Different %*

CCBs 2 041 679 17.6 2 443 949 21.5 402 270 3.9ABAs 2 612 911 22.5 2 724 264 24.0 111 353 1.5Antithrombotic 879 437 7.6 998 226 8.8 118 789 1.2Vasodilating 1 147 826 9.9 1 084 639 9.6 -63 187 -0.3Cardiac 151 016 1.3 67 903 0.6 -83 113 -0.7RAASI 2 308 527 19.9 2 131 444 18.8 -177 083 -1.1Others 1 138 075 9.8 991 794 8.7 -146 281 -1.1Antilipidemic 458 289 4.0 320 496 2.8 -137 793 -1.2Diuretics 875 156 7.5 595 560 5.2 -279 596 -2.3

*% of change after the action = (Different number/Post-action number) ¥ 100%.CCBs, calcium channel blocking agents; ABAs, alfa- and beta-adrenergic blocking agents; RAASI,rennin-angiotensin-aldosterone system inhibitors; ATC, Anatomical Therapeutic Chemical classifi-cation.

Table 3 Comparison of the difference in themedicine used numbers

Agents


Volume % Volume % Different %*

Exposure groupCCBs 260 339 17.2 293 259 20.5 32 920 3.3ABAs 337 571 22.3 348 014 24.3 10 443 2.1Antithrombotic 126 771 8.4 133 269 9.3 6 498 1.0Cardiac 22 195 1.5 9 071 0.63 -13 124 -0.8RAASI 317 173 20.9 284 841 19.9 -32 332 -1.0Others 148 815 9.8 126 782 8.9 -22 033 -1.0Antilipidemic 48 919 3.2 32 082 2.2 -16 837 -1.0Vasodilating 172 528 11.4 147 378 10.3 -25 150 -1.1Diuretics 81 441 5.4 55 137 3.9 -26 304 -1.5Total 1 515 752 100 1 430 532 100 -85 220 -5.9

Comparison groupCCBs 158 888 20.0 204 082 24.1 45 194 4.1ABAs 156 353 19.7 183 221 21.6 26 868 2.0Antithrombotic 73 047 9.2 82 653 9.8 9 606 0.6Cardiac 7 610 1.0 4 352 0.5 -3 258 -0.4RAASI 143 961 18.1 153 653 18.1 9 692 0.0Others 81 059 10.2 76 470 9.0 -4 589 -1.2Antilipidemic 30 227 3.8 22 343 2.6 -7 884 -1.2Vasodilating 74 267 9.3 70 346 8.1 -3 921 -1.0Diuretics 70 356 8.8 50 997 6.0 -19 359 -2.8Total 793 670 100 846 813 100 53 143 6.3

*% of change after the action = (Different number/Post-action number) ¥ 100%.CCBs, calcium channel blocking agents; ABAs, alfa- and beta-adrenergic blocking agents; RAASI,rennin-angiotensin-aldosterone system inhibitors.



The individual hospital GB programme implemented by theNHI in Taiwan reversed the trend of increasing rates of prescrip-tion drug costs in hospitals. The increase in the volume of medi-cines prescribed was significantly larger for the comparison groupfor both alfa- and beta-adrenergic blocking agents (1.5%) andcalcium channel blocking agents (3.9%). However, the individualhospital GB action significantly reduced the average volume ofitems on each prescription.

DiscussionIn 1998, 4 years after the introduction of NHI in Taiwan, theannual insurance premium income was insufficient to covermedical payment expenditures. The increase in medical paymentsvastly outpaced that of premium income received by the insuranceplan [9]. Because the financial balance needed to be urgentlyaddressed and constrained, a GB system was chosen as the suitableoption believed to be able to resolve the crisis. A GB systemallowed for setting an upper limit for the amount expended bymedical providers. In Taiwan, a medical payment system wascreated in which the consumer co-payment and the medicalservice provider, via prior negotiations, decided on the total healthinsurance medical service expenditure for a specific group ofmedical services. These medical services included dental care andtraditional Chinese remedies, primary care clinics and hospitalservices for a pre-determined period of time, usually a year [10]. Inaddition to the effects of the hospital GB resource allocation,quality of medical services and providers’ behaviour receivedincreasing attention [11]. However, research on the efficacy andinfluence of the GB in Taiwan was only done in the early stages ofevaluation. The study analysed changes in prescribing over the first3 years, which reflected the effects of the implementation of theindividual hospital GB system.

In Taiwan, the reimbursement rates for pharmaceuticals coveredby the NHI have always been set by the NHI. There is no variationin reimbursement rates across geographic locations or health carefacilities, and all variation across time is as a result of the BNHIreimbursement policy. NHI conducted a drug price survey in 2002,and decided to lower the reimbursement of 581 drugs, effectiveNovember 2004. We did not include the first few months immedi-ately following the action because hospitals may not show theirplanned reactions with respect to the policy change because of drugstock on hand or their previous contracts with pharmaceuticalcompanies. In 2004, NHI launched the hospital excellence program(HEP), a programme providing financial incentives to hospitals tolimit outpatient services below a certain percentage of the totalrevenue (e.g. 45% for medical centers). A total of 225 hospitalsjoined HEP while the remaining 348 hospitals did not. Our sam-pling method included patients from the same hospitals during thestudy period; the main purpose was to reduce database administra-tive difficulties and bias towards the NHI. The selected measuresused in our study were intended to capture the average drug utili-zation behaviour of the HT group associated with implantationof the NHI individual hospital GB programme. Finally, becausethe strategies used by each hospital in response to the hospital GBsystem were different, this study only represents the combinedeffects of the hospital GB.After implementation of the NHI hospitalGB system in Taiwan, in order to maintain an acceptable level ofmedical care and to avoid financial imbalances resulting in difficul-

ties in hospital operations, hospitals were forced to make adjust-ments to their management model and services provided. Althoughmodifications of the prescription policies of each hospital afterjoining the programme were unclear, it seems that hospitals joiningHEP reduced the treatment intensity of outpatient services.

Consistent with prior literature, the average number of drugitems per prescription increased for HT and diabetes year afteryear [12,13]. Chou and colleagues [12] reported that the hospitalGB effect on doctor prescriptions for the hypertension groupshowed a gradual decreasing trend after the 9th quarter post-GB.However, the number of drugs per prescription fluctuated slightlyaround 1.89 items per prescription. This result confirms that,after the action, doctors switched their prescription drugs to otherless expensive anti-hypertensive drugs or fewer items per pre-scription. Liu and Wang studied prescription patterns and timetrends for anti-hypertensive medication in newly diagnosed casesof uncomplicated hypertension in Taiwan, showing that calciumchannel blockers and beta-blockers were the most frequently pre-scribed anti-hypertensive drugs, either alone or in combinations[14]. Although diuretics were the least expensive drug option, theprescription rates for diuretics were low, at 8.3% for mono-therapies and 19.9% overall. The prescription rate for angio-tensin receptor blockers became elevated considerably over time.Chiang et al. analysed trends in anti-hypertensive drugs used bydiabetic outpatients in Taiwan over a 7-year period (1997–2003)[15]. They reported that the anti-hypertensive drugs and CCBswere the most widely prescribed class throughout the studyperiod, but that the prescribing rates declined considerably overthe study period. A significant downward trend was alsoobserved for beta-blockers and other drug classes. Drugs actingon the rennin-angiotensin-aldosterone system were the only classshowing a significant increase in prescribing rates over time.The prescribing patterns for monotherapy regimens decreasedover time, while those for two-, three-, and four- or more drugregimens increased over time. Monotherapies maintained withCCBs, beta-blockers, diuretics, and other drug classes steadilydeclined, but those maintained with drugs acting on the RASmarkedly increased. Our results reported a reduction in anti-hypertensive daily drug expense and average number of itemson each prescription after implementation of the hospital GBprogramme. Retrospective claims data provide a potentiallyuseful resource for health economics and health outcomesresearch [16–18]. However, certain limitations of our study mustbe addressed. For example, claims data are constructed princi-pally for reimbursement purposes. As a result, financial incen-tives influence the patient diagnoses included on the medicalclaim. This fact could change our average cost per patient butnot total costs. Total costs were obtained mainly from primarysources and official accounting records measuring resourcesspent that were independent of the number of patients treated.Only our estimations of cost of other drugs and the cost per visitin primary care depend on the prevalence rate. Secondly, ourhospital medication cost figures are a conservative estimationof the real cost of treating hypertensive patients. Therefore, hos-pital pharmaceutical drug costs related to chronic complicationsand additional hospitalizations of hypertensive patients may behigher than our figures. A third limitation is that claims data arenon-experimental, and as a result, additional care must be takenin inferring treatment effects [19–21].



ConclusionsStudy results coincided with our expectation that the doctorsservice at individual GB programme hospitals would switchpatients’ drugs to other less expensive drugs, and we also observeda reduction in the more expensive anti-hypertensive drugs beingused by doctors in hospitals who did not join the hospital GBprogramme. The total drug expenditure increased for both patientgroups, and the doctors service at individual GB programme hos-pitals increased at a lesser rate (-5.9%) than those of doctors’groups in hospitals that did not join the hospital GB pro-gramme(+6.3%).

AcknowledgementsThis study is based in part on data from the National HealthInsurance Research Database provided by the Bureau of NationalHealth Insurance, Department of Health and managed by theNational Health Research Institutes. The interpretation and con-clusions contained herein do not represent those of the Bureau ofNational Health Insurance, Department of Health or NationalHealth Research Institutes. We would like to thank the Tri-serviceGeneral Hospital in Taiwan for providing financial support for thisstudy (grant TSGH-C96-14-S06).

References1. Chen, T. M. (2003) Taiwan’s new National Health Insurance program:

genesis and experience so far. Health Affairs, 22, 61–76.2. Liu, S. Z. & Romeis, J. C. (2003) Assessing the effect of Taiwan’s

outpatient prescription drug copayment policy in the elderly. MedicalCare, 41, 1331–1342.

3. Insurance RoCBoNH (2004) NHI Summary Statistics. Taiwan: Insur-ance RoCBoNH.

4. Chen, T. L. (2003) Analysis on trends for health care expenditures inTaiwan. Taiwan Journal of Public Health, 21, 157–163.

5. Lee, Y. C., Yang, M. C., Huang, Y. T., Liu, C. H. & Chen, S. B. (2006)Impacts of cost containment strategies on pharmaceutical expendituresof the National Health Insurance in Taiwan, 1996–2003. Pharmaco-economics, 24, 891–902.

6. Haaijer-Ruskamp, F. M. & Hemminki, E. (1993) The social aspects ofdrug use. Finland: WHO Regional Publications, Report No. 45.

7. Tugwell, P., Robinson, V., Grimshaw, J. & Santesso, N. (2006) Sys-tematic reviews and knowledge translation. Bulletin of the WorldHealth Organization, 84 (8), 643–651.

8. Shaw, B., Cheater, F., Baker, R., et al. (2005) Tailored interventions toovercome identified barriers to change: effects on professional practiceand health care outcomes. Cochrane Database of Systematic Reviews,20 (3), CD005470.

9. Zheng, W. H. (2002) Summary of the Discussions on the FinancialProblems Facing National Health Insurance. Taiwan: National HealthResearch Institute.

10. Lin, Y. J. (2001) A Revolution in National Health Insurance PaymentPolicies – The Global Budget System. Taiwan: NPF Backgrounder.

11. Chou, Y. J., Yip, W. C., Lee, C. H., Huang, N., Sun, Y. P. & Chang,H. J. (2003) Impact of separating drug prescribing and dispensing onprovider behaviour: Taiwan’s experience. Health Policy Plan, 18,316–332.

12. Chou, C. C., Hu, K. Y., Wu, N. R., Cheng, Y. H., Loh, C. H. & Yeh, M.K. (2008) Changes in drug prescription utilization for diabetic andhypertensive outpatients after initiation of the National Health Insur-ance’s Global Budget Program in Taiwan. Medical Science Monitor,14 (5), 33–39.

13. Insurance RoCBoNH (2006) Quality Report of Hospital GlobalBudget, Q3, 2005. Taiwan: Insurance RoCBoNH.

14. Liu, P. H. & Wang, J. D. (2008) Antihypertensive medication prescrip-tion patterns and time trends for newly-diagnosed uncomplicatedhypertension patients in Taiwan. BMC Health Services Research, 18(8), 133.

15. Chiang, C. W., Chen, C.Y., Chiu, H. F., Wu, H. L. & Yang, C.Y. (2007)Trends in the use of antihypertensive drugs by outpatients with diabe-tes in Taiwan, 1997–2003. Pharmacoepidemiology and Drug Safety,16 (4), 412–421.

16. Chen, C. L., Chen, L. & Yang, W. C. (2008) The influences of Tai-wan’s generic grouping price policy on drug prices and expenditures:evidence from analysing the consumption of the three most-usedclasses of cardiovascular drugs. BMC Public Health, 12 (8), 118.

17. Schneeweiss, S., Glynn, R. J., Maclure, M., Dormuth, C. & Soumerai,S. B. (2002) Outcomes of reference pricing for angiotensin-converting-inhibitors. New England Journal of Medicine, 346, 822–829.

18. Chen, T. S. (1985) Hospital cost control in Norway: a decade’s expe-rience with prospective payment. Public Health Reports, 100, 406–417.

19. Chen, T. Z. (2003) Prevalence of Hypertension, Hyperlipidemia,Hyperglycosemia in Taiwan Area. Taiwan: Bureau of Health Promo-tion, Department of Health.

20. Anonymous (1999) World Health Organization-International Societyof Hypertension Guidelines for the Management of Hypertension.Guidelines Subcommittee. Journal of Hypertension, 17 (2), 151–183.

21. Anonymous (1997) The Sixth Report of the Joint National Committeeon prevention, detection, evaluation, and treatment of high blood pres-sure. Archives of Internal Medicine, 157 (21), 2413–2446.



Documents

Doctor pharmaceutical utilization behaviour changed by the global budget programme strategies on hypertensive outpatient prescription