Discussion
The present study aimed to evaluate the process indicators regarding the commissioned implementation of a T2D primary prevention intervention program in routine conditions of PHC, in order to facilitate subsequent analysis of effects and their interpretation and to inform future optimization of the implementation in clinical practice. In short, the main results of this process evaluation showed that the implemented strategy, primarily based on classical passive strategies of dissemination such as training and provision of resources, attained modest-to-good process indicators related to adoption, reach, and implementation of the intervention program. However, there was evidence of a notable variability in process indicators by center.
As underlined by the RE-AIM framework, one of the major factors determining the potential impact of programs is that they must be appealing to health care providers and realistic to implement in practice settings. The literature points to competing demands and lack of time as the main reasons for refusal to participate among PHC professionals [
25]. In the present study, rates of adoption of clinicians within centers were fair but improvable, especially among family physicians. In five out of seven centers in each of the comparison groups, the program adoption rate has exceeded 50 % among nurses, and only in two centers in each of the groups was the level of collaboration similarly high among family physicians. We are unable to compare these adoption rates with the findings of other studies on the translation of diabetes prevention programs as they do not report these types of data regarding the rate of adoption among health care professionals.
Overall, the reach of the screening procedures to identify individuals at risk of developing T2D among 45- to 70-year-old PHC users was 6.2 % (range 2 to 20 %).
Though these figures could be considered low, they must be considered valuable as the program has been conducted in the real context of primary health care delivery process without altering usual working conditions. All patients of that age strata who attended the centers due to any possible health complication were potential participants and were addressed by professionals after resolving the problem that caused their attendance. Other studies aimed at translating a T2D primary prevention program to PHC have attained higher reach indicators ranging from 44 to 56 %, but after conducting a previous selection of patients from electronic health records and contacting them by letter or telephone [
26‐
28]. The present study covered a much larger and broader target population than similar translation studies of T2D primary prevention in PHC, without a previous selection procedure of eligible patients being carried out, thus better reproducing the real-world conditions of PHC services. Nevertheless, between-center variability in attained reach shows that figures could be improved and this warrants further investigation on how to achieve a higher but tolerable reach.
Around half of the 45- to 70-year-old patients attending PHC who were assessed were detected as having a high risk of developing T2D (FINDRISC score ≥14), a figure within the range previously reported in the literature [
29]. However, as occurred with the reach of the screening there was substantial within-group center variability in the efficiency in detecting high-risk patients. Though all centers were instructed to approach patients and screen considering only patient age (i.e., “all patients attending the health center aged between 45 and 70 years”), some centers targeted the screening to high-risk patients, while others employed more universal strategies. These different procedures affected both reach rates and efficiency in detecting individuals who were at high risk of developing T2D: the more massive screening strategies had higher reach among attending users but lower percentages of detection of high-risk patients, the opposite occurring in centers that restricted the screening strategy by only assessing those with additional risk factors.
Not all the patients detected as having a high risk of developing T2D were finally included in the study. The research-related operational and ethical requirements, and the real-world context of PHC together with an evidenced variability in procedures and strategies adopted between centers, have determined two major issues that could affect future analysis and interpretation of the program’s effects: differences in refusal rates and in baseline comparability among comparison groups. First, a higher rate of refusal to participate in the study was encountered by collaborating clinicians in the intervention group centers. This may be attributable to the fact that these patients were not only asked to consent to the annual testing but also were invited to attend four educational workshop sessions. Second, after exclusion of a considerable proportion of undiagnosed cases of T2D (approximately 12 % in both groups), a higher proportion of women were included in the intervention group, pointing to a possible selection bias. These two sources of potential confounding would threaten the internal validity of a phase III clinical trial under controlled conditions [
30]. In the case of a phase IV implementation trial, aiming to conduct a broad evaluation of the translation of proven efficacy interventions into routine care, the focus should be on external validity and generalizability, thus assessing effectiveness in heterogeneous, unselected populations and real-world clinical settings [
30,
31]. In any case, and taking the worst case scenario of analyzing this clustered trial as an observational study, rigorous analytic procedures will need to be used in the analysis of effects of this intervention in order to reduce and deal with the possible confounding and context-specific variability [
32].
When analyzing the source of observed differences in refusal rates and in the proportion of women included, three important factors must be noted. First, as previously observed with the screening reach and its efficiency at detecting high-risk patients, refusal rate is associated with procedures and strategies for recruitment. Specifically, centers that used massive screening strategies and those that did not obtain patient consent to participate directly after the screening (offering an additional consultation) had higher odds of refusal. Second, strong between-center variability was present even after adjusting for the effect of group, these differences between centers being at least as relevant as those observed at group level for explaining differences in both refusal rates and percentage of women included. Lastly, considering that a passive implementation strategy reduces the possibility of a standardized implementation of programs, leaving the responsibility for program organization decision-making to health professionals themselves, this observed variability at center level could be expected. Variability in program adoption by professionals among different sites and in program reach and implementation indicators have also been reported in other implementation trials [
33,
34]. Every center has a different context and thus implementation strategies must be targeted and adapted to that specific context. Even after a hypothetical adaptation, results and products would be context specific as changes are common when implementing interventions in practice settings [
35]. Therefore, a lesson learned from the current process evaluation is that context is crucial, this meaning that, when using passive strategies for implementation, an a priori standardized intervention would be implemented in different ways in different settings resulting in different process and possibly clinical outcomes [
33,
36].
Evidence has clearly demonstrated that T2D can be prevented or delayed in at-risk individuals through intensive healthy lifestyle counseling [
7‐
10]. The reach of the intervention program of 48 % out of all the eligible pre-diabetic patients identified can be considered as moderate [
37]. Moreover, the quality of the intervention program implementation can be considered good, as more than 80 % of the patients in the intervention group finally included received all the components of the educational intervention program. This overall program attendance rate is within the range of other diabetes primary prevention programs in PHC [
8,
37,
38]. Again, program completion rates varied between intervention centers but it was not associated with the flexibility in the schedules and formats offered by centers.
To the best of our knowledge, this is the first study assessing the main process indicators of a translation of a T2D prevention program in the routine PHC context. However, the present study has several limitations. First, the process evaluation refers to 14 selected centers. Although collaborating centers seem to be quite diverse, they may not be representative of all PHC centers across our health service. In addition, though we have described some characteristics regarding representativeness, in terms of PHC size and composition, socio-economic status of attended populations, and so on, a lack of some other types of information regarding specific context characteristics may limit the interpretation of results. Though the external validity of the study could be questioned, implementation trials aim to conduct a broad evaluation of the translation of proven efficacy interventions into routine care, thus assessing results in heterogeneous, unselected populations and real-world clinical settings [
30,
31]. Another limitation of the present study is that it is mainly based on quantitative data. Ascertaining the factors behind implementation heterogeneity from a qualitative perspective might help in the interpretation of results.
Acknowledgements
The PreDE Group
Amara Berri PHC: Carmen Maturana Plaza, Carmen Mendinueta Maider de Miguel Sanz, Amaia Páez Viciana, María Lierni Linacisoro, María Pilar Sáenz, José Miguel Rengel, María José Salazar, Ángeles Asenjo, Coro Semperena, Esperanza Garde, María Asunción Aguirre, María Míguez, Mercedes Coloma, Nora Bergera.
Aranbizkarra I PHC: Isabel González Alonso, Concepción Isasmendi, Mercedes Torrecilla, Adela Soto, Marisol Areta, Enma Felipe, Mariví Cuervo, Ainhoa González, Flor Arbosa, Javier García y Nerea Gutiérrez.
Azpeitia PHC: María Jesús Arratibel Aguirre, Naiara Aramburu Sánchez, Irati Iza Arnaiz, Arantxa Azcune, Elena Lizaso, Janire Iparraguirre, Lourdes Goenaga, Ana Errasti, María Luisa García, Miren Elixabet Loyola, Amaia Arruti, Ascensión Benítez, Imanol Domínguez, José Luis Badiola, María Jesús Bernaras, Pedro Vidaror.
Bolueta PHC: Julia Huertas Fernández, Idoia Herrero Montiaga, Felix Menéndez Pérez, Judit Oribe Ruiz, Iñigo Alkiza, María Ángeles López Careaga, Alberto López González, Esperanza Diez, Nagore Arce.
Dumboa PHC: Benedicta Rodríguez, Juana María Álava Urrejola, Elena Erviti, Ana Cabrerizo, Eva Quintana, Lourdes De Juan, María Ochoteco, Arantza Inda, Judit Becerra, Carmen López, Itxaso Idoiaga, Isabel Roca, Maite Monge y Ana Isabel Cabrerizo.
Galdakao PHC: Marta Pérez Gómez, Mario Dolores Delgado Ansó, Mario del Camino Alegre Martínez, Marta Hierro Idirin, Pilar Etxebarria Villegas, Juan Ignacio Salgado Sáenz, Mónica Prieto Echevarría, Pilar Calvo Aedo, Agurtzane Paskual.
Irun PHC: Milagros Aramburu Beltrán, Rocío Arina Chasco, Esteban Sampedro Martínez, Oihana Saiz Alzugarai, Rosa Oronoz Lizargarte, Maisabel Maya Goya, Beatriz Larrinaga Liñero.
Iztieta PHC: Maria Angeles Sola Gaínza, Marian Ganzarain, Soledad Asenjo, Remedios Oyarzun, Rosa Espina, Onintza Aranzadi, Maddi Etxabe, Araceli de Gracia, Gemma Díaz, Laura Balague, Nuria Gañan, Eider Alberdi, Mercedes Gil, Rosa Salaverría, Mercedes Alvarez, Ana Guinea, Mikel Bagues.
Legazpia PHC: María José Lopetegui, Ramón Areizaga, Asun García, Olatz Laxalt, Lourdes Zubeldia, Idoia Garmendia, Sandra Maniegai, Izaskun Bengoa, Jon Amilibia,Ana Belén Pinilla, María Mercedes Alba, Carmen Balmaseda, Juan Miguel Urrutia, Maite Murua, Izaskun Fernández.
Llodio PHC: Fernando Fernández de Larrea Ortiz de Zarate, Begoña Huguet García, Ana Astobiza Sautua, Begoña Díaz Rodríguez, Isabel Torroba Balda.
San Martin PHC: María Asunción Fernández Rubio, María Ángeles Sedano Garate, Gemma Ayesa Pascual, María Isabel Fernández Molina, María Mercedes Martínez de Albeniz Arriaran, Justina Tobarra, María Luz Andrés Visos, María Aránzazu Elejalde López de Vergara, Manuela Vila Gómez, Estíbaliz Martínez –Alcocer
Vidal, Irina Rabalo Navarro, María Teresa Gómez Merino, José Luis Sedo Berastain, Lucia Pereda López, María Teresa Amiano Arregui, María Isabel Ibáñez López, Aurora Alcalde Lozano, María Mercedes Arroniz Perosanz.
San Vicente PHC: Amaia García Arechavala, Adelina García Roldan, Elba García Martínez, María Begoña Bao Brea, María Teresa López Torres, M Del Castañar Sánchez Gómez, Pedro Ángel Martínez Rama, Ana Isabel Pereda García-Ola, Carmen Sánchez Gutiérrez, Encarnación Santamaría Pelarda, M Jesús Aragón Miranda, Mercedes García Asenjo, Tomas Larrazabal Intxausti.
Txurdinaga PHC: Teresa Rodríguez González, Beatriz Navascues Cepeda, Belén Sangroniz Mateo, Pilar Bustamante Castaño, Begoña Nieves Etxeberria, Julián Sedano, Begoña Aguirre Larrondo, Arrate Orive, Sara de la Fuente Mediavilla, María Teresa Bárcena de Miguel, Naike Galán Fie, Purificación De Cos Rodríguez, Carmen Abalo Duran, Iratxe Gancedo, María Mar García Ramírez, Esther Etxegarai, Concepción Arizaga Valer, Eugenia Barrenengoa Pérez, Jesús María Ruiz Barrio, Javier Orbegozo Urcelay, José Ángel Fernández Gallo.
Zuazo PHC: María Jesús Galarza, Carmen Ayuso, Karmele Benito del Valle, Floreal Crespo, María Díaz de Rada, Miriam Pozas, María Felisa Loyola, Florencia Martín Gómez, María Luisa Rodríguez Ortiz de Zarate, María Consuelo Verdes, Nieves Ortego, Elena Resines, María Isabel Lázaro, Yolanda Villalba, Belén Gutiérrez Pañales, Josune López, Ana Cristina Fernández González, Begoña Arias, María Teresa Rodeño, Jon Burgos, Alicia Hormaetxea, Madalena Presmanes Medi, Begoña Ayerdi, María José Cordovilla, Concepción Estébanez.
Osarean: Raquel Roca Castro, Inmaculada Nido Menchaca, Roberto Sanchez Robredo, Maite Blanco Rodriguez, Nagore Fernandez Ruiz, Laura Gancedo Mendez, Jon Koldobika Gil Borica, Virginia Martin Ramos, Laura Gil de Muro Arce, Olatz Hurtado Lugera, Miriam Lafuente Vegas, María Estrella Muñoz Caso, Nagore Lejarza Seguin, Elena Maestre Ania, Amaia Martinez Martin, Ana Cristina Acero Duro, Alexander Alonso Fradejas, María Asuncion Calvo Goitia, Sonia Del Rio Martinez.
Coordinator Group: Alvaro Sánchez, Carmen Silvestre, Natalia Campo, Naroa Valderrama Andikoetxea, Catalina Martinez Carazo, Gonzalo Grandes, Regina Sauto.
The project was supported by the Basque Government Department of Health (EXP: 2010111039; IC10-05), Kronikgune Research Institute (EXP: KRONIK 11/033), and by the Carlos III Institute of Health of the Spanish Ministry of Health, and co-financed by European Union’s ERDF (FIS 2012/00495; RETICS G03/170 and RD06/0018/0018; CAI08/01/0065).