Factors associated with the diffusion rate of innovations: a pilot study from the perspective of the Brazilian Unified National Health System

Budget impact analyses require a set of essential information on health technology innovation, including expected rates of adoption. There is an absence of studies investigating trends, magnitude of budgetary effects and determinants of diffusion rates for health technology innovations worldwide during the last decades. The present study proposes a pilot assessment on main determinants influencing diffusion rates of pharmaceutical innovations within the Brazilian Unified National Health System (SUS). Data from the Brazilian Health Informatics Department (DATASUS) was gathered to establish the main determinants of diffusion rates of health technology innovations in Brazil, specifically referring to pharmaceutical innovations incorporated in the Brazilian Program for Specialized Pharmaceutical Services (CEAF) at SUS. Information was retrieved on DATASUS relating to patients who had used one of the medicines incorporated into CEAF at least three years prior to the beginning of the study (2015) for treatment of each health condition available. Thus, data from patients adopting 10 different medicines were analyzed in the study. Results from the zero-one inflated beta model showed a higher influence on diffusion rates of pharmaceutical innovations due to: number of pharmaceutical competitors for treatment of the same disease available at CEAF (negative); medicine used in combination with other medication (positive); and innovative medicine within the SUS (positive). Further research on diffusion rates of health technology innovations is required, including wider scope of diseases and medications, potential confusion factors and other variables that may influence rates of adoption in different health systems. Diffusion of Innovation; Biomedical Technology Assessment; Health Evaluation Correspondence F. M. Sarti Universidade de São Paulo. Av. Arlindo Bettio 1000, São Paulo, SP 03828-000, Brasil. flamori@usp.br 1 Departamento de Assistência Farmacêutica, Ministério da Saúde, Brasília, Brasil. 2 Universidade de São Paulo, São Paulo, Brasil. 3 Fundação Instituto de Pesquisas Econômicas, São Paulo, Brasil. 4 Departamento de Gestão e Incorporação de Tecnologias, Ministério da Saúde, Brasília, Brasil. Schneiders RE et al. 2 Cad. Saúde Pública, Rio de Janeiro, 32(9):e00067516, set, 2016 Introduction The objectives of the Brazilian Unified National Health System (SUS) include assurance of universal health care coverage and integral health assistance within a publicly financed health structure. Nevertheless, recent demographic and epidemiologic trends in Brazil, along with rapid developments of technological innovations, have been posing challenges to health system management. Numerous technical alternatives available for adoption in health care have been producing continuous increases of health expenditures 1. Policies towards health technology assessment may support rational incorporation of innovations in national health systems to guarantee economic sustainability. In 2011, the National Committee for Health Technology Incorporation (CONITEC) was established to support the Brazilian Ministry of Health on decisionmaking processes related to health technology assessments and the incorporation of therapeutic innovations 2. CONITEC is responsible for issuing recommendations on health innovations to be incorporated, excluded or modified within the SUS, supported by scientific evidence on efficacy, accuracy, effectiveness, safety and costs; including economic evaluation and budget impact analysis (BIA), from the SUS perspective 3. There has been growing interest in BIA recently 4 and Health Technology Assessment (HTA) agencies in several countries request BIA to support decision making processes on the adoption of health technology innovations 5,6,7,8. Economic evaluation studies provide useful information on the adoption of innovations; however, their results lack information on potential economic impacts on national health accounts. BIA results include the projection of expenditures due to the incorporation of health technology innovations for diagnosis and treatment of populations during specific periods, based on a comparison of alternative scenarios, using payer perspectives 1,9,10. Yet, in order to perform BIA, essential information on health technology implementation is required: prices, prescription, and adoption rates. Estimates of adoption rates in health systems are usually based on studies of diffusion rates of technological innovations; which are influenced by population characteristics, including communication among individuals (e.g., prescription, marketing, or patients’ requests) and predisposition for technology adoption (e.g., physicians’ or patients’ preferences for innovation); however, there are controversies regarding the magnitude of effect from diverse variables 5,11,12,13,14,15,16. Consequently, information to perform BIA is usually based on market-specific evidence or experts’ consultations. Delay in technology adoption after incorporation to the health system may occur due to differences in personal characteristics among diverse health professionals and patients (e.g. resistance in acceptance of innovations, or lack of information), changing budget impacts over time; thus, diffusion rates of health innovations are crucial for BIA 4. This study proposes an innovative approach to identifying factors influencing diffusion rates of medicines incorporated within the SUS, in order to provide evidence to support further advances in health technology assessment.


Introduction
The objectives of the Brazilian Unified National Health System (SUS) include assurance of universal health care coverage and integral health assistance within a publicly financed health structure. Nevertheless, recent demographic and epidemiologic trends in Brazil, along with rapid developments of technological innovations, have been posing challenges to health system management. Numerous technical alternatives available for adoption in health care have been producing continuous increases of health expenditures 1 .
Policies towards health technology assessment may support rational incorporation of innovations in national health systems to guarantee economic sustainability. In 2011, the National Committee for Health Technology Incorporation (CONITEC) was established to support the Brazilian Ministry of Health on decisionmaking processes related to health technology assessments and the incorporation of therapeutic innovations 2 .
CONITEC is responsible for issuing recommendations on health innovations to be incorporated, excluded or modified within the SUS, supported by scientific evidence on efficacy, accuracy, effectiveness, safety and costs; including economic evaluation and budget impact analysis (BIA), from the SUS perspective 3 . There has been growing interest in BIA recently 4 and Health Technology Assessment (HTA) agencies in several countries request BIA to support decision making processes on the adoption of health technology innovations 5,6,7,8 .
Economic evaluation studies provide useful information on the adoption of innovations; however, their results lack information on potential economic impacts on national health accounts. BIA results include the projection of expenditures due to the incorporation of health technology innovations for diagnosis and treatment of populations during specific periods, based on a comparison of alternative scenarios, using payer perspectives 1,9,10 .
Yet, in order to perform BIA, essential information on health technology implementation is required: prices, prescription, and adoption rates. Estimates of adoption rates in health systems are usually based on studies of diffusion rates of technological innovations; which are influenced by population characteristics, including communication among individuals (e.g., prescription, marketing, or patients' requests) and predisposition for technology adoption (e.g., physicians' or patients' preferences for innovation); however, there are controversies regarding the magnitude of effect from diverse variables 5,11,12,13,14,15,16 . Consequently, information to perform BIA is usually based on market-specific evidence or experts' consultations.
Delay in technology adoption after incorporation to the health system may occur due to differences in personal characteristics among diverse health professionals and patients (e.g. resistance in acceptance of innovations, or lack of information), changing budget impacts over time; thus, diffusion rates of health innovations are crucial for BIA 4 .
This study proposes an innovative approach to identifying factors influencing diffusion rates of medicines incorporated within the SUS, in order to provide evidence to support further advances in health technology assessment.

Methods
Detailed data from the Ambulatory Information System (SIA) of the Brazilian Health Informatics Department (DATASUS) was gathered to establish determinants of diffusion rates of health technology innovations in Brazil, specifically referring to pharmaceutical innovations incorporated in the Brazilian Program for Specialized Pharmaceutical Services (CEAF). This option was selected on account of the availability of information on patients using numerous types of medication for treatment of diverse health conditions, available online within the DATASUS platform using TabWin software (DATASUS. http://portal. saude.gov.br/portal/se/datasus/area.cfm?id_ area=732).
CEAF databases encompassed nationwide information on pharmaceutical services, reported by Brazilian states to the Ministry of Health, referring to every event of medication distribution for each patient (identified using the cryptographic number from the National Health Card), diagnosis, and medication characteristics. Data available from any patients using any medicines incorporated into CEAF at least three years before the beginning of this study (2015) for treatment of any health condition was included in the analysis, ensuing data from patients adopting 10 medicines (Table 1).
Considering evidence from the literature 15,16 , a dataset containing 17 categories of variables potentially associated with adoption rates of medication within the SUS was generated, including variables related to characteristics of diseases, respective medications and its prices, treatment protocols and costs (Table 2).
Diffusion rates were based on the percentage of patients using medication among patients Cad  Verifies the influence of need to adopt a combined use of medication, due to potential difficulties to access other medicines prescribed.

Method
Analysis of the first PCDT available for the targeted disease, in order to identify indication of use in association with other medicines.

Innovation within the SUS Binary variable (Yes, No) Description
Analyzes the influence of incremental benefits of the pharmaceutical innovation in comparison to other types of treatment of the disease.

Method
Due to absence of specific definition regarding the concept of innovation in health care, the following premises were adopted: Medication for treatment of diseases not yet available at SUS; Medication for treatment of diseases already available that: Represents new line of treatment of the disease; or Presents improved efficacy in relation to other medication already available, based on search of evidences published in meta-analysis or direct comparison 20,21 .
Other medications competing in the same line of treatment and in the same pharmacological category were not considered innovative.

Time gap between from incorporation and clinical protocol publication (months) Discrete variable (Count) Description
Analyzes the influence of PCDT in diffusion rates, due to definition of prescription and utilization criteria.

Method
Identification of the publication date of PCDT. If the PCDT was published prior to the medication incorporation, the variable value was zero.  Description Analyzes the influence of the area of medical specialty of the disease on diffusion rates of pharmaceutical innovations.

Method
Analysis of the PCDT for the targeted disease, in order to determine the area of medical specialty for treatment of the disease. Each disease was categorized in only one specialist area, if more than one area was indicated; the most representative specialist area was adopted. 14. State of residence of patient Binary variable for each

Brazilian state Description
Verifies differences among states of residence of patients in the access of medication provided by SUS or in execution of CEAF, and its influence on diffusion rates.

Method
Extraction of data regarding patients' state of residence from SUS databases, described in Methods.
(continues) Cad. Saúde Pública, Rio de Janeiro, 32(9):e00067516, set, 2016 Verifies differences among regions of residence of patients in the access of medication provided by SUS or in execution of CEAF, and its influence on diffusion rates.

Method
Extraction of data regarding patients' region of residence from SUS databases, described in

Methods.
16. Long-term use medication Binary variable Analyzes the influence of period recommended for treatment on diffusion rate, considering that continuous-use medication usually presents lower adherence from patients.

Method
Analysis of general recommendations regarding the period recommended for treatment using the medication in the PCDT for the targeted disease, at the period of pharmaceutical innovation incorporation within SUS. Long-term use medication was considered to be indicated for utilization during periods longer than one year of treatment. In the case of pharmaceutical innovations without published PCDT at the moment of incorporation, information contained in recent PCDT were adopted.

Results
Considering 17 categories of independent variables described, seven categories presented association with diffusion rates (Table 3). Results from the zero-one inflated beta model showed a higher influence on diffusion rates of pharmaceutical innovations due to: the number of pharmaceutical competitors for treatment of disease available at CEAF (negative); medicine used in association (positive); and innovative medicine within the SUS (positive).
Variables related to the characteristics of pharmaceutical innovations were prominent to diffusion rates within the SUS, whereas organizational characteristics of the health system adopting innovations were mostly represented by region of residence of patients, which may account for major differences in infrastructure and management of the SUS.
There was a set of variables without a statistically significant association with diffusion of pharmaceutical innovations within the SUS: other treatments available at CEAF; line of treatment; time gap to clinical protocol publication; type of disease and medical specialty; patent; price in comparison to competitors; management level responsible for acquisition; patients' state of residence; and route of administration. Nevertheless, considering the correlation between variables excluded and some variables included in the model, it was expected that part of the variables tested would be omitted.

Discussion
There is a lack of evidence in the scientific literature regarding theoretical frameworks and methods to support BIA referring to adoption rates of health technology innovations within national health systems, either in the public or private sectors. A limited number of studies investigated trends of budgetary effects and determinants of diffusion rates of recent health technology innovations worldwide. This study proposes a pilot assessment on determinants influencing Cad. Saúde Pública, Rio de Janeiro, 32(9):e00067516, set, 2016 Table 3 Coefficients of zero-one inflated beta model for diffusion rate of pharmaceutical innovations in the Brazilian Unified National Health System (SUS). Brazil, 2015.  diffusion rates of pharmaceutical innovations within the SUS. Results obtained were consistent with prior knowledge on diffusion rates of novel medication 15,16 , showing trends for faster diffusion rates of medication with incremental benefits, and treatments with higher costs inducing higher demand within the SUS. Slower adoption rates of medication with substitutes (competitors) within CEAF indicated impact on prices due to competition in the pharmaceutical market.
This study investigated influences on the rate of diffusion of pharmaceutical innovations within the SUS on two of three possible dimensions 15 : medication features and organizational characteristics. The third dimension, regarding characteristics of individuals, was not within the scope of databases used to perform the analyses; thus, a main limitation of the study is the lack of information on the preferences of physicians, patients, society and the pharmaceutical industry.
Another limitation is the potential duplication of patients' records within DATASUS data-bases; considering evidence showing that quality of data extracted from DATASUS may be compromised due to duplication or lack of consistency in the patients' registry, depending on the type of information used 18,19 . Nevertheless, with regard to the CEAF data, limitations are considerably lower, due to the need of medical prescriptions for treatment of each patient diagnosed within the SUS during the same period (monthly).
Contributions made by this pilot study in measuring diffusion rates of pharmaceutical innovations and identifying their immediate determinants should be acknowledged due to internal validity of analysis and viability for reproduction to support further evidence for BIA research.
In order to ensure the external validity of results obtained in this study, further research on diffusion rates of health technology innovations is required, including a wider scope of diseases and medication, potential confusion factors and other variables that may influence rates of adoption in different health systems. Cad. Saúde Pública, Rio de Janeiro, 32 (9)  The journal has been informed about some errors in the paper. The corrections are follows: A revista foi informada sobre alguns erros no artigo. As correções seguem abaixo: La revista fue informada sobre algunos errores en el artículo. Siguen las correcciones:

refiriéndose a las innovaciones farmacéuticas incorporadas en el Programa brasileño para Servicios Farmacéuticos Especializados (CEAF) en el SUS. La información fue rescatada del DATASUS relativa a pacientes que habían usado una de las medicines incorporadas al CEAF al menos 3 años antes del comienzo del estudio (2015) para tratamiento de cada condición de salud disponible. Así, fueron analizados datos de pacientes que usaron 10 medicamentos diferentes. Los resultados del modelo de regresión beta aumentada mostraron una influencia más alta en las tasas de difusión de las innovaciones farmacéuticas debido a: número de competidores para el tratamiento de la misma enfermedad disponible en el CEAF (negativo); medicamentos usados en combinación con otra medicación (positivo); y medicina innovadora en el SUS (positivo). Se requiere más investigación adicional sobre las tasas de difusión en tecnología de la salud, incluyendo un enfoque más amplio de las enfermedades y su medicación, potenciales factores de confusión y otras variables que quizás influencien las tasas de incorporación a los dife
• Where the text read: This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited.    Other medications competing in the same line of treatment and in the same pharmacological category were not considered innovative.

Time gap between from incorporation and clinical protocol publication (months) Discrete variable (Count) Description
Analyzes the influence of PCDT in diffusion rates, due to definition of prescription and utilization criteria.

Method
Identification of the publication date of PCDT. If the PCDT was published prior to the medication incorporation, the variable value was zero.  Description Analyzes the influence of the area of medical specialty of the disease on diffusion rates of pharmaceutical innovations.

Method
Analysis of the PCDT for the targeted disease, in order to determine the area of medical specialty for treatment of the disease. Each disease was categorized in only one specialist area, if more than one area was indicated; the most representative specialist area was adopted.  Analyzes the influence of period recommended for treatment on diffusion rate, considering that continuous-use medication usually presents lower adherence from patients.

Method
Analysis of general recommendations regarding the period recommended for treatment using the medication in the PCDT for the targeted disease, at the period of pharmaceutical innovation incorporation within SUS. Long-term use medication was considered to be indicated for utilization during periods longer than one year of treatment. In the case of pharmaceutical innovations without published PCDT at the moment of incorporation, information contained in recent PCDT were adopted.   Verifies the influence of need to adopt a combined use of medication, due to potential difficulties to access other medicines prescribed.

Method
Analysis of the first PCDT available for the targeted disease, in order to identify indication of use in association with other medicines.

Innovation within the SUS Binary variable (Yes, No) Description
Analyzes the influence of incremental benefits of the pharmaceutical innovation in comparison to other types of treatment of the disease.

Method
Due to absence of specific definition regarding the concept of innovation in health care, the following premises were adopted: • Medication for treatment of diseases not yet available at SUS; • Medication for treatment of diseases already available that:  Analyzes the interference of drug therapy costs per patient in the diffusion rate and potential impacts of reduction in prices due to scale in production, considering that overall budget impact may influence the access to medication within the SUS.   Analyzes the influence of period recommended for treatment on diffusion rate, considering that continuous-use medication usually presents lower adherence from patients.

Method
Analysis of general recommendations regarding the period recommended for treatment using the medication in the PCDT for the targeted disease, at the period of pharmaceutical innovation incorporation within SUS. Long-term use medication was considered to be indicated for utilization during periods longer than one year of treatment. In the case of pharmaceutical innovations without published PCDT at the moment of incorporation, information contained in recent PCDT were adopted.