Pr>cm>ocoE~I(O~~ ' , lW5 1 11 ()- I ~SIJl . SJIO
ORIGINAL RESEARCH ARTICLE
The Nelson Prescribing Project A Programmed Intervention in General Practice in New Zealand Rutll I. Fergllsofl,1 Clare E. Salmolllfl and Timothy ,.B. Maling] Department of Medicine, Wellington School of Medicine, Wellington, New Zea land 2 Department of Public Health, Wellington School of Medicine, Wellington, New Zealand
Summary
We have defined the effect and acceptability of a locally developed general practice progmmmc for the modification of prescribing. This volunlary programme consisted of prescription analysis and feedback. followed by visits from a pharmacist, a therapeutic bulletin on benzodiazcpine prescribing. and use of a locally compiled preferred medicines list. A 3 -month prescription sample from 26 gencml practitioners (GPs) fu lfill ing a stable practice definition was used to compare prescribing pre-project and mid-project. For 20 out of 26 GPs, prescribing of medicines on the preferred medicines list had increased significantly 8 months after the intelllention programme had been introduced. Total prescription numbers and total medicines expenditure dec reased by 8.3 and 4.9%, respectively, from 1988 to 1989. The decrease in bcnzodiazepine prescribing was marked (mean - 22.2%. range - 50.3 to +4%). The cooperative multi model approach was highly successful in modifying prescribing in general pract ice.
The quality of prescribing. particularly with respect to efficacy, safety and value for money, is of considerable interest in New Zealand,II -31 as in all developed countries. Escalation of expenditure on mcdicinesl41 in New Zealand has fuelled fears of legislation to lim it prescribing. The national m anageme nt of pharmaceul icals in New Zealand has been directed primarily towa rds pricing policies. Computerised pricing of prescribed pharmaceuticals by the New Zealand Department of Health was not introduced until April 199 1, and prior reliance on a ma nual prescription pricing system had lim ited the avai labil ity of prescribing information in New Zealand. In the Nelson prov ince in New Zealand, general practitio ners (G Ps) who were members of a primary healthcare development group identifi ed pharmaceuticals as a target for cost savings. An intervention
programme to modify GP prescribing was subsequently established in conjunction with the academ ic clinical phannacology unit of the Wellington School of Medicine. and supported by the Nelson Area Health Board. The aim was to test the efficacy, acceptability and cost effectiveness of complemenlary strategies to modify prescribing in general prac tice. These included a Preferred Medicines List (PML), personalised audi t re ports on prescribi ng. visits from a pharmacist and targe ted medicines information. This study provided the blueprin t for the New Zealand Preferred Medici nes Concept. 151
Methods The Nelson provi nce is situated in the northwest of the South Island of New Zealand. The population of 7 1 400 is concent rated in one urban region (5 1.1 %) and in scal1ered horticultural/agricultural
Ftrgusoll et ~I.
556
sell1cmcnlS on the Waimea Plains and in Golden Bay. The rati o of GPs per head o f popul ation is hi g h compared with o ther agriculturally-based provincial New Zealand regions,l61 Prescribing Intewentions 44 oul o f a possib le 45 GPs in the Nelson region participated in the stud y. A PML was prepared by local GPs and supponed by acl inica] pharmacologist (TJB Maling), fo llowing recommendations from regional workshops that were sponsored by the lo cal subfaculty of the
Royal New Zealand College of General Practitioners (RNZCG P). The criteria for selection ora medicine for the PM L were efficacy. safety, convenience of use and COSI. The draft PML was c irculated to all GPs in the Nelson re gion. and it was et sled for 3 months ( February to Apri l 1989) prior to further rev iew and preparation of the first ed iti on of the PML. Throughout the interventio n period (February to October 1989), reminder notices were sent to alt GPs with all laboratory reports to re inforce usc of the PML. Confidential prescribing profiles were prepared for each GP for the calendar months of February and October 1989.1') The format was developed through consultation with the prescribers.!8) A dual currency system (number of limes prescribed and cost) was used to prese nt information in graphic and tabulated fo rmats for both individual GPs and for the group as a whole. A special section also displayed informatio n o n ind ividual and grou p compliance with the PML. Each G P was visited by the project pharmacist in A ugust 1989 and February 1990. after receipt of their prescribing reports. The visiting pharmacist's fun ction was to assist with report interpretation and to identify funher information requirements from the prescribing re port da tabase o r from literalUre sources. Following the identification of a high regional level o f bcn zodiazepine prescribing identified in 1988. targeted medic ines information was provided to all the G Ps in the Nelson area in the stud y.
This infornUlIio n was provided at the same time as the rest of the intervention programme was put into place. This consisted of a meeting arranged through the Nel son RNZCGP subfaculty and di stribution of a therapeuti c bulletin on the treatment of insomnia and anxiety. Agraphical representation of bcn7.odiazepine prescribing as a atio r of total prescribing. with each recipient identified in relation t o t heir colleagues. was included in the prescription audit feedback. In the PML. temazepam was specifically substituted for the popular hi storical pre ference for triazolam.
Outcome Measurements Prescription Audit Prescriptio ns presented to the New Zealand Departme nt o f Health Prici ng Office from pharmac ies in the Nelson area were used to evaluate the influence of the project o n prescribi ng patterns. The sampling procedure was validated through the use of self-inking duplicale prescription pads during one of the sample months (October 1989). Aeo unt of the form s re ce ived by the Prici ng Office was compared with a count of the duplicate copies. The retrieval rate was a mean of93% (range 86 to 10 0%). indicating t he validity of the sampli ng procedure. Parlicipation was limited to GPs in fuJI -t ime practice. in the same practice for at least 3years prior to the project, and not absent from the practice for more than 3 weeks per sampling period. Prescribing was compared between two 3-monlh sampl es (A ug ust to October), pre-project (1988) and midproject ( 1989) . Prescriptions were identifi ed by prescribing dale and the prescriber's s ignature. Data were entered onto a Paradox 3 computer programme. therapeutic class and PML status were appended, and then the prescription was priced. Forcost-comparative purposes. a base pricing method was used and the presc riptions inboth years were priced using the 1989 New Zealand Drug Tariff. Data were tran sferred to SAS file s (SAS Institute Inc., North Carolina, USA) for the statistical analysis o f prescribing c hanges for both individual GPs and the pooled sample. Analysis of PML compliance in-
The Nelson Prescribing Project
557
Table I. Total number 01 prescriptions. according to maiOr therapeutic class. wrinen by 26 general practitioners. luHil li ng a s table practice delinitiof1 wIIo participated in ttle Nelson Presclitling Project. over two 3-mooth periods (August to October). in t988 (before introdllClion 01 an intervention programme designed to modify prescribing panems; pre·project) and 1 yearlaterin 1989 (mid·project). Wilcoxon ranked sum test statistics (z, p) for interpractitioner variation between 1988 and 1989 are shown Therapeutic class
Total number 01 prescriptions
"68
Cardiovascular Systemic antibiotic Dermatological Psychiatric
,'" 7275
''''
"" ""
z-Value
1989 7013
"'3 "" 4407
-1.94 -2.06 --4 .17
-<.'"
Musculoskeletal
4161
Gastrointestinal
2929
"" "" "" 3142
....92 -1.61 - 1.45
Gynaecological
2638
2132
... 80
Respiratory Neurological
5218
clud.ed the 218 drugs or drug combinations in the PML booklet, and the medicines that were speci fically excl uded. Medicines that cou ld only be prescribed on the recommendation of a speciali st (c.g. cancer chemotherapy) were excluded from the analysis. Participant Evaluation
Participants' evaluation of the project was assessed by aquestionnaire mai led in April 1990. A 100% response was obtained. Project features were evaluated using several modalities, including a IOcm visual analogue scale (VAS), descriptive choice and direct q uestioning. The responses were analysed using aMacintosh Stat view program.
Results Number of Prescriptions Written
Of the 44 GPs who participated in the study. only 26 fulfilled the stable practice definition. The number of prescriptions written per GP decreased significanlly between 1988 (pre-project) and 1989 (mid-project) [mean -8.3%, r.mge -24.9 to +18.1%; t25 = 2.99; P = 0.006], although an increase occurred for 6 GPs. The volume of medicines per prescriplion and the number of days of treatment supplied per prescription did not differ significantly between the 2 periods (p = 0.91 and p = 0. 11 , respect ively, log transformed data). CI Adis InlemoliQrxJI Umlled . M rights rewrved.
-0.97
p-Value
0.00 0'" <0.001 <0.001 0.33 "O.OOt
0.10 0.14
All major therapeutic classes, except gastrointestinal and systemic antibiotics, showed are· duction in pre scription numbers from 1988 to 1989 (table I). Reinstatement of laxatives onto the New Zealand Drug TarifTand an atypically high incidence of respiratory infections were important extraneous infl uences during the project intervention period. The greatest reduction in prescription numbers was for psychiatric medicines, particul arly benzodiazepines (mean decrease - 22.2%, range - 50.3 to +4%). Other than the notable but minor effect of the market imroduction of zopiclone in 1989 (to represent 6.4% of hypnosedativc prescriptions), thi s trend was not matched by an increase in prescribing of therapeutic alternatives such as non-benzodiazepine hypnosedatives (-16. 1%) and tricyclic antidepressants (- 11 .3%). Other classes showing a major decrease in prescription numbers were dennatological (-18. 1 %) and neurological preparations [dominatcd by analgesics (- 19.2%)J, and musculoskeletal medicines ldominated by nonsteroidal anti-inflammatories (-6.3%) 1. The Preferred Medicines List and Prescribing Choice
Medicines covered by the PML accou nted for 98.5% of all the prescriptions in 1989. The increase in the proportion of preferred medic ines prescribed after the introduction of the programme was statistically significant for 20 out of 26 GPs
Ferguson el al.
558
5.0 to 145.6 medicines; prange 0.026 to <0.(01). The overall mean change was +5.8% (range - 2. 1 t o +13.4%). From the 1988 baseli ne of5 [ 349 prescrip· tions written, the reduction in the number of prescri ptio ns written during the 1989 mid-project study period was greater for the medicines not listed on the PM L (-3994) than for medicines listed on the PM L (-222). There was a IO- fold reducti on in prescribing expenditure for non- PML-listed med icines (-$NZ45 889) compared wit h PMLlisted medicines (-$NZ4493) in 1989 dollars; the total expenditure in 1988 was $NZI 029619. Changes in benzodiazepi nc prescribing showed considerable variabi lity between GPs (table II). Temazepam. the PML hypnotic choice, was prescribed by GPs more frequently in the 1989 sample compared with the 1988 sample after the first audit report. the benzodiazepine bu lleti n and the seminar (+21.6%); this contrasted with a decrease in prescribing of triazoJam (-40.0%), the historical hypnotic choice. The substitution of temazepam for triazo[am was significant (t25 ",4.25, p 0.3). The Cost of Prescribing
The reduction in the number of prescriptions written duri ng the interventi on in [989 translated to a reduction in expenditure of 4.9% ($NZ50 382 over 3months). However, in [2 practices, the mean ex pendi ture per prescription increased sig nificantly (p < 0.05, Wilcoxon tests).
Two major therapeutic classes, systemic antibiotics and respiratory medicines, did not follow the general relationship between change in expenditure and change in prescription numbers. The increased expenditure on systemic antibiotics (+[4.5 %) substanti ally exceeded the increase in the number of dose uni ts (+8.5%). The enhanced popu larity of the comparatively expensive ant ibiotic combination amoxicillin/clavu[anic acid. from 11.9% of all systemic anti biot ic p rescriptions in [988 to 16.6% in 1989, was a major factor. The number of prescriptions for respiratory medicines declined over the study period, but expenditure increased by $NZ8290. Thi s was attributable i n pari to an increase in prescribing of 2 recently introduced fonn u[ations. a PML-listed budesonide aerosol (+$NZ8226), an unl isted sodium cromoglycate (cromolyn sodium) 5mg aerosol (+$NZ 1946), and a c hange in government policy of a reinstatement on the New Zealand Drug Tariff of laxati ves (+$NZ9952) and hi stamine H I-receptor antagonists (+$NZ2334). Participants' Questionnaire
All of the GPs except one descri bed the project as useful (26) or of significant use ( 16), and one GP did not respond. The rating of the 4project features were stati stically different (Friedman chi- square) '" 12.77, P = 0.005). On a VAS scale of zero (no value) to 1.0 (greatest val ue), audit reports (median 0.74) and the visit ing pharmacist (median 0.71) were rated more highly tha n the PML (med ian 0.66) or medicines infonnation (median 0.56). Participation in meetings to discuss medicines choice for the PML did not significantly influence the frequency with which GPsconsulted the PM L, or their assessment of its value. Perceived modification
Table II. Summary 01 the change in number of benzodiazepine prescriptions for the 26 general practitioners in a stable practice situation. participating in the Nelson Prescribing Project. over the mid·season sample (August to October) before (1988) and after (1989) the project intervention Statistic
AJ I benzodiazepnes
Triaz~am
Mean
-'"
_H
.,
--<5
Standard deviation Lowes! Highest
23
-
C> AdiJ lnternollonol LnVIed AI rights """"rved.
"
-H
Temazepam
Temazepam:triaz~am
.3
to.52
-".3"
-0.26
ratio
U"
"" PhormocoEconomics 7 (6) 1995
559
The Nelson Prescribing Project
Medicines expenditure New Zealand
Nelson province
20
g
, ."
O)--tf-,.==!"It:::::-::::- -~----------------------------- I
Di spensing numbers
~
u
New Zealand Nelson province
-20
""
".,
""
Year
""
""
""
Fig . 1. Change 'rom t986 in dispensing numbers and medicines expenditure (in actual dollar values but exduding dispensing lees) IOf New Zealand and the Nelson province.
of benzodiazepine prescribing att itude (yes/no) was linked by prescri bers to their assessment of the targeted benzodiazepine informat ion (Wilcoxon z :::0 2.92, p :::0 0.003 ) and with their attendance at a benzod iazepine discussion meeting (Fisher exact test, p:::o 0.084).
Discussion The change in prescribing observed using the longitudinal sampling technique is strongly suggestive of a project-related change, but it does not discount the possibility of ex ternal influences produc ing similar trends. The lack of good quality prescribing data in New Zealand means that this si sue cannot be readily resolved. Until the introduction of computerisation in 1991, only total dispensing count and total tariff cost data were collected by the prescription p ricing service of the Departmen t of Heailh. Since 1986, prescribing trends for Nelson and New Zealand have been parallel, except for expenditure during the project period and the subsequent year (fi g. I ). The major drop in di spensing numbers in 1989 indicates the effect of a change in the o Adis IoTerrool\onol l.iTVTe
system o f medicines di stribution fo ll owing the introduction o f a further increase in the patient prescri ption charge. Medicines in New Zealand are traditionall y supplied for a maximum period of 3 months and arc dispensed in monthly i nstalments. After the introduction of a fee per prescription in J 985 and subsequent increases in 1989 and 1991. items costing less than the prescription fee could be purchased directly by the patient. The increase in the prescription item fee in 1989 did not significantly affect the supply or filling of prescriptions costing less than the fee in the study sample (- 1.8%, standard deviation 3.5). When the medicines expendit ure f igures fo r 1988 and 1989 are compared, the 9.3 % decrease in expenditure in the Nelson province is not mirrored by the national figures, which show a 2.3% increase (fig. I). In 1989. a change in the governmentregul ated wholesale margin had a significant effect on inflatio n. The Fi shers Ideal Index l91 calculated from the August to October 1988 and 1989 prescriptions for the study sample was 0.980, indicating a small decline in item cost. PhormocoEcOl"lOf"l'ics 7 (6) 1995
560
The dominance of regional prescribing by proj~ eel participants (8 1%) and the reduction in postproject Nel son prescriplion expenditure. but not national expenditure. slrongly suggests a projcctrelated effeci. The inc rease in the prescription charge in 1989 clearly influen ced dispensing numbers, but any effeci on the quantity of med icines prescribed could not be asce naincd as such data were not collected nationally. However. the small increase in national medicines expenditure and the slable infl ation TalC would suggest re lalive volume stabi lity nati onall y. This contrasts with the project stud y sample, in which a significant decline in actual medicines volume was observed . The decline in triazolam prescri bing and the adoption of the PML choice temazepam is counter to national prescribing trends (Department of Health . personal communicatio n. 1990). and provides further evidence for the o perational effectiveness of the PM L. Ho wever. the marked reduction in total triazolam prescription num bers cOlTesponded with a modest increase in te mazepam pre scriptions. The substantial decli ne in benzodiazepine prescribing suggested the potential for simple focu sed interventions for modifying prescribing . e specially for Ihose therapcUlic si sues perceived as controversial and sensiti ve. Evaluations o f pilot programmes for modifying general practice prescribing are infrequently reported in the literature . General practice formul aries are widely used in the UK,I IO.II ) but their specific effects o n med ic ines c ho ice have been poorly investigated. Interventions thaI mimic the pharmaceutical industry and that have been used t o rationalise medicines choi cel12.13 1 or improve cost efficiency have been modestly successful. Feedback on prescribi ng has bee n shown to be innuential, but the e ffec t is poorly sustained when fe edback is discontinued.II"1 With the possible exception of locally developed fo rmularies. all other interve ntions have con fo rmed to an impositional model. developed outside the targ et community and introd uced by coercive or mandatory means. In Nel son. a more autonomous approach was chosen ; GPs were invited 10 playa key role in evaluating and managing their
Ferguson cl al.
o wn prescribing. Abasic feature of the oc ncept was the use o fthe PML as a prescribing benchmark,l51 with the di scussions o n m edicines selection providing an important educational focu s. Personal prescribing audits. together with the provision of medicines infonnation. provided an information base fo r rational dec ision making . In add ilion to the provision of infonnation. the role of the visiting phannac istl71 may have imparted an impression of scrutiny and accountabil ity. but in a manner acceptable to the project participants. The Nelson Prescribing Project has been successful in introduci ng a medici nes manageme nt programme which is considered beneficial by the G Ps. and which can modify medicines choice and the cost of prescribing. These res ults provided the justification for the development o f t he New Zealand Preferred Medicines Concept,l 5) launched in August 1991.
Acknowledgements We wish t oacknowledge the assistance of Dr Bob Boyd and Mr Warren Thompson of the Ne w Zealand Department of Health. Funding was provided by grams from Ihe Ikpan· menl of Health. lhe New Zealand Lotteri es Board and the Heahh Research Council ofNe w Zealand.
References I. Malcolm LA. Melli A. Ream trends indrug preK ribing r:lles and Co:sts in New « aland . N Z Med J 1988: 10 1: 233·6 2. Isherwood J. Malcolm LA. Hornblow AR. Fac tors associa ted with vari ation in gc nc..~ 1 practitioner prescri bing eo:su. N Z Med J 1982: 95: 14·7 3. Sinelair BL. Ashton T. Jackson R. e ta1. Tre nds in antihypl'rte n. sive """di cation COSts in a oohan of Auc kl andcrs . N Z Med J 1989: 102: 52 1·3 4. SUllon F. Trends i n pharmaceutica l eXpI'nditure over the last decade . Wellin gton: Depanmc nt of Hea lth. 1988 5. Maling TJ B. The New Zealand Pre ferred Medic ines Co ncept a natio nal sc he""" for a udit and q ualit y assu.... nee o fp rescriJ>. ing . l'Ilarrrw;oEconomics 1994; 6 ( I); 5· 14 6. Barnell JR . Changes in the geographic diS!ributio n of general pracl itioners in New Zealand since 1975. I: Regiooal and urban· ru ral d iffe rences. N Z Med J 199 1; 1(l.I: 3 t4·5 7. n,rguson RI . Maling TJR. 1lIe Nelson General Practice Pre· scribing Project. Pan I: I pilot aud it of regional prescribing profiles. N Z Med J 1990: 103: 558-60 8. Frrguson RI . Mating TJB . The Nelson General Practice Pre· K ribing Project. Pan 2:prescribi ng repons for se lf aud il. N Z Med J 1990: 103: S60-2 9. Forsyth FG. Fowler RF. 1'hcory and practice of c hai n prj« in · de~ nu mbers. J R Soc Stal 1981 : 144: 224·6 10. Drury M. Practice formu laries. Practitioner 1990; 234: m·5
FhomoooEconorrics 7 (/») 1995
561
The Nelson Prescribing Project
II. Grant GR, Gregory DA, yan Zwanenberg TD. De,·e lopment of
a limited formulary for general practice. Lancet 1985: I: 1030-2 12. Soomerai SB, AYorn J. Principles of educational outreach ('ac· ademic detailing·) to improve clinical decision making. JAMA 1990: 263: 549-56 13. Landgren Fr. Harvey KJ. Mashford ML. et al. Changing antibioti c prescribing by educat ional marketing. Mcd J Aust 1988: 149: 595-9
14. Harris CM. Fry J. Jarman B, ct al. Prescribing - a case for prolonged treatment. J R Coli ~n Pract 1985: 35: 284·7
Correspondence and reprin ts: Associale P rofessor Timothy
Ma/irlg, Department of Medicine, Wellington School of Medicine, PO Box 7343, Wellington South, New Zealand.
PhorrrocoEcOl"lOl"lll<::o 7 (6)
1~