Costs of Public Pharmaceutical Services in Rio de Janeiro Compared to Farmácia Popular Program

ABSTRACT OBJECTIVE To analyze the costs of public pharmaceutical services compared to Farmácia Popular Program (Popular Pharmacy Program). METHODS Comparison between prices paid by Aqui Tem Farmácia Popular Program (Farmácia Popular is available here) with the full costs of medicine provision by the Municipal Health Department of Rio de Janeiro. The comparison comprised 25 medicines supplied by both the municipal pharmaceutical service and Aqui Tem Farmácia Popular Program. Calculating the cost per pharmaceutical unit of each medicine included expenditure by Municipal Health Department of Rio de Janeiro with procurement (price), logistics, and local dispensation. The reference price of medicines paid by Aqui Tem Farmácia Popular was taken from the Brazilian Ministry of Health standard in force in 2012. Comparisons included full reference price; reference price minus 10.0% copayment by users; and maximum reference paid by the Ministry of Health (minus copayment and taxes). Simulations were carried out of the differences between the costs of Municipal Health Department of Rio de Janeiro with the common medicines and those potentially incurred based on the reference price of Aqui Tem Farmácia Popular. RESULTS The Municipal Health Department of Rio de Janeiro spent R$28,526,526.57 with 25 medicines of the common list in 2012; 58.7% accounted for direct procurement costs. The estimated costs of the Health Department were generally lower than the reference prices of the Aqui Tem Farmácia Popular Program for 20 medicines, regardless of reference prices. The potential costs incurred by Health Department if expenditure of its consumption pattern were based on the reference prices of Aqui Tem Farmácia Popular would be R$124,170,777.76, considering the best scenario of payment by the Brazilian Ministry of Health (90.0% of the reference price, minus taxes). CONCLUSIONS The difference in costs between public provision by Municipal Health Department of Rio de Janeiro and Farmácia Popular Program indicates that some reference prices could be reviewed aiming at their reduction.


INTRODUCTION
Difficulties in access to medicines due to their high share in private household spending in Brazil 16 and supply problems in public units of the Brazilian Unified Health System (SUS) 2 justified the introduction of Farmácia Popular Program (PFPB) by the Brazilian Ministry of Health (MH) in 2004 13 . Its creation, however, did not alter the responsibilities of municipalities in the provision of medicines from SUS.
The program expanded into arrangements involving the public and private sectors. It has its own network of pharmacies and a public-private partnership with the retail pharmaceutical sector, both of them with or without copayment by users.
The number of pharmacies accredited by the private division of PFPB, called Aqui Tem Farmácia Popular, increased over 750% in 2006-2013 and was responsible for the program's geographical spread 15 . It has its own rules of operation and includes 41 products to treat the most prevalent diseases in the population, most of which feature in the basic list provided by pharmacies from the SUS network. A list of antihypertensive, antidiabetic and antiasthmatic medicines is exempt from co-financing 3,13,15 .
The Ministry of Health's direct disbursement to the accredited retail network is based on a reference price (RP) per medicine. This price is valid for the entire country and comprises, in addition to the procurement price, costs related to logistics, dispensation and maintenance of pharmacies, insurance and taxes.
The costs of the ATFP Program are subject to questioning within the public model for provision of access. An audit by Tribunal de Contas da União (TCU -Federal Court of Accounts) raised the debate on the differences in prices found in public tenders compared to those in the Program. The audit identified a huge discrepancy between the RP paid by ATFP and the prices of 13 medicines purchased by the public sector in April 2010. For four of them, the difference between the RP and the average procurement price exceeded 1,000% a . TCU pointed out that simple comparison with bidding prices was not enough to establish whether one program is more efficient than the other, and therefore studies considering other costs involved are necessary.
New medicine supply arrangements in Brazil are important strategies to back policies aimed at expanding access. However, they lack data to support analyses of funding sustainability and more efficient use of public resources.
This study aimed to analyze the costs of public pharmaceutical services compared to Farmácia Popular Program.

METHODS
A comparative study of costs (in reais -R$) of the two models of pharmaceutical services (PS), the ATFP Program and public services, in Rio de Janeiro, Southeastern Brazil, in 2012.
Secretaria Municipal de Saúde do Rio de Janeiro (SMS-RJ -Municipal Health Department of Rio de Janeiro) has a recognized track record of efficiency in the procurement of medicines b and was one of the first in Brazil to make purchases using the price registration system 8,20 . Calculating the cost per pharmaceutical unit (PU) of each medicine included SMS-RJ costs with procurement, logistics (storage, distribution and transport) and local dispensation. Administrative costs involved in procurement or loss and misplacement were not computed.
Procurement costs were obtained from the price registration minutes of SMS-RJ bidding processes, published in the Municipal Official Gazette, following the methodology used by Silva and Caetano 14 . The publication features the unit price for each item.
The source of logistics costs was the SMS-RJ outsourcing contract for this activity. The amount paid to the company was prorated to all medicines, considering the following specific adjustments: (i) monthly consumption of each medicine in PU; (ii) storage and transport space (in m 3 stored and shipped); (iii) labor force employed in separating medicines, considering the number of items per order; and (iv) insurance, considering the average value in stock. This allowed isolating the contribution of each medicine of the common list between municipal public provision and the ATFP Program, enabling the individualization of a price per PU linked to logistics costs.
Dispensation costs involved expenses related to human resources of basic network pharmaceutical services; purchase of materials, furniture and equipment; and building management and maintenance services. The result of this cost component was subsequently prorated to isolate the costs with the PU of each medicine.
Staff costs were based on a specific SMS-RJ census of September 2012, which identified the number of personnel dedicated to pharmacy, separated by professional category -pharmaceutical and support staff (pharmacy technician, administrative personnel, among others) -and employment relationship (direct municipal administration and SO). Civil servant salaries were determined from the municipal administration salary scale related to the average monthly remuneration values of the corresponding levels, including hazard pay, three-year service bonus and transportation allowance. The salaries of professionals hired by SO were obtained from the SMS-RJ internal system ( An apportionment method was applied to the administrative costs of service contracts and maintenance, which cannot be individually measured by service item. Physical area (m 2 ) was used for energy, water and sewage, cleaning/maintenance, security and building maintenance; quantity of equipment for air conditioning and refrigeration; and number of network access points in pharmacies and number of logins to access Internet service providers and electronic medical record systems, respectively.  Dispensation costs per PU were estimated according to the formula below. The percentage share of each product in the common list was calculated from the total expenditure of SMS-RJ with the procurement of basic medicines in 2012. This percentage share was applied to the total dispensation cost and weighted by the annual consumption of each medicine, obtained from the municipal Medicine Distribution Center.
Where  Based on the annual consumption of SMS-RJ, simulations were performed on the differences between SMS costs with each medication of the common list and those potentially incurred based on the RP of ATFP.
Excel © software was used to estimate costs and comparisons.
The study was approved by the Ethics Committees of the Institute of Social Medicine -Rio de Janeiro State University (Opinion 170,617/2012) and of SMS-RJ (Opinion 350A/2012). There was significant variation in the percentage share of cost components, especially for some specific products (Figure). For most medicines, procurement costs were the  highest, accounting for more than 70.0% of costs in the following products: norethindrone; estradiol + norethisterone; NPH and regular human insulin; beclomethasone (250 mcg); budesonide and benserazide + levodopa.
Comparing the costs of SMS-RJ and the RP of ATFP Program, municipal costs were generally lower, regardless of the RP considered ( Table 2). The difference between full RP and cost per PU in SMS-RJ was 279.8% higher. With the 10% copayment discount, variations were slightly lower on average: 270.7%. In the most favorable comparison with ATFP (minus copayment and taxes), the percentage difference was 3.5 times lower than the federal program, on average.
Captopril and enalapril showed percentage differences above 1,000%, with a 14 times lower cost in the municipal context, regardless of the comparison scenario with ATFP. Five other medicines also stood out favorably in the municipal dispensation, with costs at least five times lower than ATFP, even after discounting copayment and taxes: alendronate sodium, losartan, glibenclamide, atenolol and ipratropium solution ( Table 2).  Janeiro); PU: pharmaceutical unit; RP: reference price a Simulation using full reference price paid by ATFP, disregarding copyament percentages. b Simulation using reference price minus 10% copayment by users. c Simulation using maximum reference price paid by MH minus taxes.
Simulations of potential costs incurred by SMS-RJ if its medicine consumption pattern were paid according to the RP of ATFP showed values of R$124,170,777.76, considering the best payment scenario by MH (90,0% of RP minus taxes). SMS-RJ would spend 3.4 times more than estimated with the entire common list if it used the RP of ATFP (Table 3).

DISCUSSION
Comparison between the reference price paid to establishments accredited by the Brazilian Ministry of Health in the ATFP Program and the costs of public supply of common list medications by SMS-RJ, calculated per pharmaceutical unit, in general showed differences in favor of the municipal public service, which were 3.5 times greater, on average. The cost of municipal provision was lower in 20 of the 25 items in common with the ATFP Program. Considering the demand of each drug consumed in 2012 by SMS-RJ, the simulation showed that if the municipality had acquired them for the lowest RP, it would incur over R$95 million in the overall cost for the same 25 products.
The ATFP Program offers medicines fully covered by CBAF for use in primary health care. Nationwide expenditure with ATFP totaled R$1,293,874,112.05 in 2012, of which approximately R$261 million were used exclusively to pay pharmacies established in Rio de Janeiro, RJ i . This volume of resources reinforces the importance of analyzing such expenditures vis-à-vis those of other agencies, especially municipal ones, which bear the heavier burden in providing pharmaceutical services related to primary care.
For most products, procurement costs were the highest compared to the other components investigated. A common feature of medicines with procurement costs above 70.0% of the total cost per PU was low volume purchases.
The contraceptives levonorgestrel and estradiol + norethisterone are acquired centrally by MH and passed on to the municipality, with low volume purchases by SMS. A similar condition occurs with insulin, which is only purchased by the municipality when the MH supply is disrupted. Beclomethasone and budesonide, indicated for treatment of asthma and rhinitis, have little market competition due to the low number of products and manufacturers in the country and the absence of generics, and were only incorporated into SMS-RJ procurement in 2011. Benserazide + levodopa had a single manufacturer until 2012, which classified it as a unique product, hindering price negotiations.
Low volume purchases and less bargaining power contribute to higher prices and inefficient medicine purchases in SUS 19 . Higher volume purchases attract greater interest from suppliers, expand competition, and are usually associated with price reductions. They tend to attract manufacturers who offer prices closer to production costs, with reduced cost per unit compared to retailers 9,11,18 . Associations or consortia of institutions for joint procurement also indicate that medicine prices are sensitive to economies of scale and negotiating power 1,5 .
Stocks improve the level of services by deploying resources needed for the production process and encouraging economies of scale in procurement, protecting against price increases and demand uncertainty 17 . Storage and warehouse operating costs should be added to the costs of medicines in stock, proportional to product specificity (cooling requirements) and physical storage area 4 .
Such elements account for the more prevalent logistics costs of medicines transferred by MH for local distribution, such as ethinyl estradiol + levonorgestrel and NPH and regular insulin, and therefore tend to occupy a lot of storage space and/or demand significant human resources for frequent monthly transfers.
Ipratropium bromide spray and carbidopa + levodopa showed proportionally higher dispensation costs compared to the others, with no clear justification observed. This is probably due to the low costs of the other components, increasing dispensation proportionately. Low volume purchases, extremely specific use instructions, and relatively low consumption demand support this hypothesis.
The differences identified in the comparisons were more favorable to costs at municipal level: SMS-RJ cost estimates resulted in prices below those of ATFP for 20 medicines. The RP paid by MH to private pharmacies is more than four times the price paid by SMS-RJ, considering procurement, logistics and dispensation costs, in eight products. These differences remain relevant even in the best payment scenario (RP minus taxes and copayment by users, when applicable).
The more advantageous results of SMS persist when using the median rather than the average of the percentage differences. The RP for medicines featured in the ATFP list are established considering the factory price approved by Câmara de Regulação do Mercado de Medicamentos (Medicine Market Regulation Chamber), information on market revenue and retail trade volume of the medicines, and the average discount on the factory price of the respective medicines 15 . The observed differences certainly raise questions about the possible overestimation of the RP of some pharmaceutical inputs, even though they are determined based on the lowest factory prices approved.
The identified differences intensify the debate raised by the 2011 TCU report. Aggregating the other cost components produced important variations. The percentage difference of 2,507.0% between the median of the municipal cost and the RP of ATFP for the medicine captopril in the abovementioned document fell to 1,325.6% in the most favorable scenario for MH. A similar reduction trend occurred with enalapril: from 1,937.0% to 1,325.6%.
Despite the comparative reductions, the RP paid by MH are still 14 times higher than prices paid by SMS. These angiotensin-converting enzyme inhibitors, recommended for treating systemic high blood pressure and congestive heart failure 10 , are widely used in SUS 6 .
A similar investigation carried out by Carraro j compared the maximum RP paid for 13 medicines available from ATFP, minus taxes, with their costs per PU estimated for 12 municipal health departments in nine states, including logistics and dispensation costs. The RP of ATFP was, on average, 150.1% higher than the estimated costs of municipal pharmacies, and no medicine price paid by MH was inferior to municipal costs.
Although the percentage differences are systematically favorable to estimated municipal costs per PU in both studies, there was a greater disparity in SMS-RJ (3.5 to 28.6 times lower). City size, volume purchases, epidemiological profile and local health system structure, including pharmaceutical services, may have contributed to these findings.
Simulations based on medicine consumption in SMS-RJ in 2012 and expenditures estimated by the lowest RP of ATFP also showed savings in municipal provision, with the difference in overall costs exceeding R$95 million if prices of the ministry program had been used. Items with a RP lower than the estimated municipal cost have low consumption and, analyzed in an integrated way, have a small impact and do not reverse the municipal advantage. The savings correspond to more than three times the amount of federal transfers (R$31,562,221.00) from CBAF to SMS-RJ.
There are other debates related the concurrence of the two pharmaceutical services provision models: the size and overlapping of lists of medicines available from SUS units and ATFP accredited retail pharmacies, which lead to service duplication; lack of use of public sector purchasing power; concentration of the Brazilian Ministry of Health as the key player in supplying medicines for basic care, which is primarily provided at municipal level 12,13 .
The advantages of municipal provision found in this study cannot, however, be generalized to other local arrangements and realities. The estimated costs concern a single location. up to 30,000 inhabitants. The size of the population and the health service network favor the volume of purchases and increase the city's power of negotiation. Coupled with the relatively well-structured management of pharmaceutical services, they can potentially reduce costs and maximize efficiency. Finally, other unobserved factors such as the complex logistics in a continental-size country and the different sales taxation schemes of states and municipalities also have repercussions.
The study was limited to a single year, providing a static picture in time. It may not have picked up some relevant cost elements of the municipal pharmaceutical chain that might have emerged in a longer study period, such as seasonality, prescription profile changes, etc.
Administrative costs directly involved in the procurement process and those from loss and misplacement were not included. The former are difficult to be individualized in the pharmaceutical services management chain, especially because the human resources involved in the administrative process of medicine procurement also engage in other procurement activities. This prevents time estimates or the definition of an apportionment criterion capable of identifying this element. Moreover, in general there is only one bidding process per year. The literature on losses is scarce and national statistics are unavailable. Medicine procurement in SMS-RJ is by electronic trading with price registration, with on-demand delivery based on consumption, enabling smaller stocks and minimizing losses from product expiry. Misplacements are covered by insurance provided in the logistics contract, which was included in the specific component costs. The health units have a technical accountability structure and dispensation rules that allow control and reduce misplacement.
This study reinforces the importance of costs in analyzing SUS policies, in particular those related to pharmaceutical services and the provision of medicines by the government. The pharmaceutical services model of the ATFP Program is based on the logic of medicine consumption as a promoter of access, with no emphasis on matters related to costs compared to large public medicine purchasers, who must consider their financial sustainability. The difference in costs between the public provision of SMS-RJ and ATFP indicates that some of the reference prices could be reduced, when compared to prices paid by SMS, which are systematically more favorable.