SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 2021, VOL. 39, NO. 3, 355–363 https://doi.org/10.1080/02813432.2021.1958497 ORIGINAL ARTICLE Classifying and communicating risks in prediabetes according to fasting glucose and/or glycated hemoglobin: PREDAPS cohort study Enrique Regidora,b,c,d, Lucıa Cea-Sorianoa,b, Antonio Ruizd,e,f, Albert Godayg,h,i, David Carabantesa, Javier Dıez-Espinod,i,j,k, Sara Artolad,l and Josep Franch-Nadald,m,n,o; for the PREDAPS Study Group a Department of Public Health and Maternal and Child Health, Faculty of Medicine, Universidad Complutense de Madrid, Madrid, Spain; bInstituto de Investigacion Sanitaria del Hospital Clınico San Carlos (IdISSC), Madrid, Spain; cCIBER Epidemiologıa y Salud Publica (CIBERESP), Madrid, Spain; dredGDPS Foundation, Madrid, Spain; eCentro de Salud Universitario Pinto, Madrid, Spain; f Universidad Europea de Madrid, Madrid, Spain; gServicio de Endocrinologıa, Hospital del Mar, IMIM, Barcelona, Spain; hDepartment of Medicine, Universitat Autonoma de Barcelona, Barcelona, Spain; iCIBER Fisiopatologıa de la Obesidad y Nutricion, Madrid, Spain; j Tafalla Health Center, Navarra, Spain; kInstituto de Investigacion Sanitaria de Navarra (IDiSNA), Pamplona, Spain; lCentro de Salud Jose Marva, Madrid, Spain; mBarcelona City Research Support Unit, University Institute for Research in Primary Care Jordi Gol, Barcelona, Spain; nCIBER Diabetes y Enfermedades Metabolicas Asociadas, Madrid, Spain; oDepartment of Medicine, Universidad de Barcelona, Barcelona, Spain ABSTRACT ARTICLE HISTORY Objective: Information about prognostic outcomes can be of great help for people with predia- Received 11 January 2021 betes and for physicians in the face of scientific controversy about the cutoff point for defining Accepted 26 April 2021 prediabetes. We aimed to estimate different prognostic outcomes in people with prediabetes. KEYWORDS Design: Prospective cohort of subjects with prediabetes according to American Diabetes Prediabetes; absolute risk; Association guidelines. odds; progression; Main Outcome measures: The probabilities of diabetes onset versus non-onset, the odds cohort study against diabetes onset, and the probability of reverting to normoglycemia according to different prediabetes categories were calculated. Results: The odds against diabetes onset ranged from 29:1 in individuals with isolated FPG of 100–109 mg/dL to 1:1 in individuals with FPG 110–125 mg/dL plus HbA1c 6.0–6.4%. The prob- ability of reversion to normoglycemia was 31.2% (95% CI 24.0–39.6) in those with isolated FPG 100–109 mg/dL and 6.2% (95% CI 1.4–10.0) in those with FPG 110–125 mg/dL plus HbA1c 6.0–6.4%. Of every 100 participants in the first group, 97 did not develop diabetes and 31 reverted to normoglycemia, while in the second group those figures were 52 and 6. Conclusions: Using odds of probabilities and absolute numbers might be useful for people with prediabetes and physicians to share decisions on potential interventions. KEY POINTS Communicating knowledge on the course of the disease to make clinical decisions is not always done appropriately.  Prediabetes is an example where risk communication is important because the prognosis of subjects with prediabetes is very heterogeneous.  Depending on fasting plasma glucose and HbA1c levels, the odds of probabilities against dia- betes onset ranged from 29: 1 to 1: 1.  Depending on fasting plasma glucose and HbA1c levels, the number of subjects in 100 who revert to normoglycemia ranged from 31 to 6.  Using probabilities and number absolutes on the prognosis of prediabetes may be useful for people with prediabetes and physicians to share decisions on potential interventions. Introduction identification of these individuals requires knowledge Identifying patients at risk of developing a disease in of the natural history of the disease and empirical evi- order to implement interventions that prevent or dence that supports the effectiveness of the interven- delay onset is a key activity of clinical practice. Early tion [1–2]. The information gleaned from this process CONTACT Lucıa Cea-Soriano tlcea@ucm.es Department of Public Health and Maternal and Child Health, Universidad Complutense de Madrid, Pza. Ramon y Cajal, s/n. Ciudad Universitaria, Madrid 28040, Spain Supplemental data for this article can be accessed here. ß 2021 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 356 E. REGIDOR ET AL. must then be communicated to patients, enabling effects of the interventions without this early identifi- them to make an informed decision about potential cation implying an improvement in their prognosis. interventions. However, communicating knowledge on The aim of this study, performed in a prospective the course of the disease is not always done appropri- cohort of people with prediabetes according to ately. Instead, individuals with a pathological or meta- American Diabetes Association (ADA) criteria, is to esti- bolic anomaly frequently hear that they are at higher mate several prognostic outcomes according to a risk for developing the disease [3–4]. This information wide range of FPG and/or HbA1c values and to show may be insufficient for making a decision, as a situ- the results as odds of probabilities and as absolute ation that doubles the risk of developing the disease numbers to aid patients and their physicians in deci- could reflect an absolute increase of 10 to 20%, or of sion-making. 0.1 to 0.2%. One option is to prioritize communication on abso- Subjects, material and methods lute risk [5–7]. This strategy may sometimes require knowledge on the magnitude of the risk of develop- Study design ing the diseases for a variety of values, for example, The Primary Health Care Study on the Evolution of when the range of abnormal metabolic values (above Patients with Prediabetes (PREDAPS Study) followed a normal but below the diagnostic threshold) is wide. cohort of people with prediabetes and another with- With this information, people situated at different out glucose metabolism disorders. Details of the study points along this range could better understand their have been described previously [19]. Family physicians chances of developing the disease. Similarly, in situa- throughout Spain conducted the study in the context tions in which metabolic indicators could revert to of everyday clinical practice. In 2012, people aged 30 normal, it is necessary to communicate the likelihood to 74 years who presented to primary health care for of this happening. any reason were invited to participate in the study. Prediabetes is a case in point. The term refers to sit- Exclusion criteria were: diabetes, terminal disease, uations in which blood glucose or glycated hemoglo- pregnancy, surgery or hospital admission in the three bin (HbA1c) is higher than normal but lower than the months prior to baseline, or any hematological disease cutoff for diabetes diagnosis [8–9]. People with predia- which could alter HbA1c values. Analyses were betes present a higher risk of developing diabetes restricted to the prediabetes cohort. Individuals were than those without [8–10], but some also revert to included in this cohort if they met ADA criteria for normoglycemia [9,11,12]. Thus, in people with differ- prediabetes: FPG levels of 5.5 to 6.9 mmol/L (100 mg/ ent levels of prediabetes, communication of this infor- dL to 125 mg/dL) and/or HbA1c range from 39 to mation could be useful for deciding between different 47 mmol/mol (5.7% to 6.4%). In 2012, 1184 people strategies to lower the risk of diabetes or raise the with prediabetes were enrolled; 831 completed the probability of returning to normal glycemic levels. five years of follow-up to 2017. Therefore, 353 subjects This informed decision-making could also be of dropped out of the study before 2017. These subjects great help to clinicians given the controversy among were included in the analyzes and contributed to the the scientific community with regard to the lower limit risk of diabetes and the likelihood of reverting to of fasting plasma glucose (FPG) and HbA1c for defin- normoglycemia up to the time of loss to follow up. ing prediabetes [13–17]. The American Diabetes Mean follow-up was 4.2 years. Association (ADA) established that a subject with FPG Participants gave their written informed consent between 100 and 125 mg/dL or with HbA1c between before inclusion. The study was classified by the 5.7% and 6.4% has prediabetes [8]. According to some Spanish Agency of Medicines and Medical Devices as authors, physicians should propose weight loss or a non-intervention (observational) post-authorization physical exercise, to delay the development of dia- study, and the study protocol was approved by the betes, to subjects with these a FPG or HbA1c values. Parc de Salut Mar Clinical Research Ethics Committee And, in some cases, incorporate pharmacological ther- in Barcelona (2011-4274-I). apy [13–15]. However, other authors consider that these values represent an unnecessary burden for the Procedure health system, since most of these subjects will not develop diabetes in at least 10 years of follow-up and After collecting baseline information about sociode- a large part will revert to normoglycemia [16–18]. In mographic characteristics, medical history, and health- addition, these subjects will experience the secondary related habits from participants’ clinical records, their SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 357 family physicians conducted personal interviews. FPG reverted to normoglycemia, and continued at predia- and HbA1c values were obtained annually. Participants betes levels were calculated (the latter was calculated were followed for up to five years or until dropout, by subtracting the sum of the two previous propor- death, or diabetes onset. tions from the unit) and represented by pictograms. Diabetes was defined as either FPG 7 mmol/L The probability of diabetes non-onset was calcu- (126 mg/dL) or HbA1c 48 mmol/mol (6.5%). lated in a complementary way as the probability of Reversion to normal glucose regulation was consid- diabetes onset. Since diabetes non-onset can be con- ered if FPG and HbA1c values were FPG < 5.5 mmol/L sidered a successful outcome, we also calculated the (100 mg/dL) and HbA1c <39 mmol/mol (5.7%), odds between the probability of diabetes non-onset respectively, at the time of the last visit. versus onset. For example, an odds of 3 (3:1) would For this study, participants were divided into eight mean that diabetes non-onset is three times more categories on the basis of baseline diagnosis of predia- likely than diabetes onset. And after converting the betes: isolated FPG 5.5-6.0 mmol/L (100–109 mg/dL), odds into probability, that is, dividing the odds by one isolated FPG 6.1–6.9 mmol/L (110–125 mg/dL), isolated and then adding the odds, it would mean that 3 out HbA1c 39–41 mmol/mol (5.7–5.9%), isolated HbA1c of 4 people would not develop diabetes. 42–47 mmol/mol (6.0–6.4%), FPG 5.5–6.0 mmol/L Finally, additional analyses were used to test the (100–109 mg/dL) plus HbA1c 39–41 mmol/mol extent to which the differences in prognosis between (5.7–5.9%), FPG 5.5–6.0 mmol/L (100–109 mg/dL) plus prediabetes categories could be due to differences in HbA1c 42–47 mmol/mol (6.0–6.4%), FPG 6.1–6.9 mmol/ health behaviors. Cox regression analysis was used to L (110–125 mg/dL) plus HbA1c 39–41 mmol/mol determine the magnitude of the association of predia- (5.7–5.9%), and FPG 6.1–6.9 mmol/L (110–125 mg/dL) betes categories with the risk of diabetes onset and plus HbA1c 42–47 mmol/mol (6.0–6.4%). with reversion to normoglycemia. The first model included adjustment for sex and age. In the second, health behaviors at baseline were added to model 1. Statistical analysis And in the third model, health behaviors assessed at For each prediabetes category, the sex- and age- each year of follow-up were entered as time-depend- adjusted incidence rate of diabetes and the sex- and ent covariates and added to model 1. age-adjusted incidence rate of regression to normogly- cemia per 1000 person-years were calculated, with Results 95% confidence intervals (CI) [20]. The standard popu- lation for adjustment (distribution by age and sex) Supplementary table 1 shows the baseline characteris- was the entire prediabetes cohort. Based on these tics—demographic distribution, biochemical parame- incidence rates, the probability of diabetes onset or ters, hypertension, obesity and health-related reverting to normoglycemia during follow-up were cal- behaviors—according to each category of prediabetes. culated. Incidence rates were assumed to remain con- Table 1 shows the number of participants at baseline, stant over five years of follow-up, such that each person-years of follow-up, and sex and age-adjusted probability has been estimated as 1 – exp ( incidence incidence rates. The overall incidence rates of diabetes rate  5) [21]. Likewise, based on these probabilities, and reversion to normoglycemia were 42.8 per 1000 for every 100 people in each prediabetes category, the person-years (95% CI 37.0–48.6) and 43.9 per 1000 number of participants who developed diabetes, person-years (95% CI 38.1–49.8), respectively. The Table 1. People at the beginning of follow-up, person-years of follow-up, and sex and age-adjusted incidence rates of diabetes and regression to normoglycemia per 1000 person-years according prediabetes category. People at the beginning of Person-years of Incidence rate Incidence rate of Prediabetes category follow-up follow-up of diabetes (95% CI) regression (95% CI) Isolated FPG 100–109 mg/dL 152 680 6.8 (0.6–13.0) 74.8 (55.0–100.9) Isolated FPG 110–125 mg/dL 102 426 38.4 (20.5–56.3) 36.2 (19.3–55.8) Isolated HbA1c 5.7–5.9% 204 941 9.9 (3.7–16.1) 84.5 (66.2–103.9) Isolated HbA1c 6.0–6.4% 112 501 19.3 (6.5–32.2) 38.7 (5.9–50.0) FPG 100-109 mg/dL & HbA1C 5.7–5.9 % 138 600 35.2 (18.6–51.8) 29.0 (16.4–43.6) FPG 100-109 mg/dL & HbA1C 6.0–6.4 % 128 524 52.8 (30.3–75.3) 23.2 (5.3–36.7) FPG 110-125 mg/dL &HbA1C 5.7–5.9 % 140 579 55.5 (36.1–74.8) 30.3 (14.6–44.1) FPG 110-125 mg/dL &HbA1C 6.0–6.4 % 208 671 132.3 (103.9–160.8) 12.9 (2.8–21.0) Overall 1184 4922 42.8 (37.0–48.6) 43.9 (38.1–49.8) 95% CI: 95% confidence interval. 358 E. REGIDOR ET AL. Table 2. Sex and age-adjusted probabilities of diabetes non-onset and onset, odds of diabetes non-onset versus onset and sex and age-adjusted probability of reversion to normoglycemia. Diabetes Probability (%) of diabetes Probability (%) of Odds of non-onset Probability (%) of reversion to Prediabetes category non-onset (95% CI) diabetes onset (95% CI) versus onset (95% Cl) normoglycemia (95% CI) Isolated FPG 100–109 mg/dL 96.7 (93.7–99.7) 3.3 (0.3–6.3) 29: 1 31.2 (24.0–39.6) Isolated FPG 110–125 mg/dL 82.5 (75.5–90.3) 17.5 (9.7–24.5) 5: 1 16.6 (9.2–24.3) Isolated HbA1c 5.7–5.9% 95.2 (92.3–98.2) 4.8 (1.8–7.7) 20: 1 34.5 (28.2–40.5) Isolated HbA1c 6.0–6.4% 90.8 (85.1–96.8) 9.2 (3.2–14.9) 10: 1 17.6 (2.9–22.1) FPG 100–109 mg/dL & HbA1C 5.7–5.9 % 83.9 (77.2–91.1) 16.1 (8.9–22.8) 5: 1 13.5 (7.9–19.6) FPG 100–109 mg/dL & HbA1C 6.0–6.4 % 76.8 (68.8–83.5) 23.2 (14.1–31.4) 3: 1 11.0 (2.6–16.8) FPG 110–125 mg/dL &HbA1C 5.7–5.9 % 75.8 (68.6–85.9) 24.2 (14.1–31.4) 3: 1 14.1 (7.0–19.8) FPG 110–125 mg/dL &HbA1C 6.0–6.4 % 51.6 (44.8–59.5) 48.4 (40.5–55.2) 1: 1 6.2 (1.4–10.0) Overall 80.7 (78.4–83.1) 19.3 (16.9–21.6) 4: 1 19.7 (17.3–22.0) 95% CI: 95% Confidence interval. PREDAPS Study. 1. Probability of diabetes non-onset versus onset. lowest incidence rate of diabetes was observed in the levels during follow-up. Of every 30 individuals with prediabetes category for isolated FPG 100–109 mg/dL; isolated FPG 100–109 mg/dL and of every 2 individuals and the highest, in individuals with FPG 110–125 mg/ in the FPG 110–125 mg/dL plus HbA1c 6.0–6.4% dL plus HbA1c 6.0–6.4%. As for reversion to normogly- group, 29 and 1 did not develop diabetes, respect- cemia, the highest rate was in the group with isolated ively. For every 100 subjects in the first group 97 indi- HbA1c 5.7–5.9%, and the lowest in the category with viduals did not develop diabetes and 31 normalized FPG 110–125 mg/dL plus HbA1c 6.0–6.4%. the glycemia, while in the second group the figures Table 2 shows the probability of diabetes non-onset were 52 and 6, respectively. versus onset, as a percentage, and the odds against Participants’ health-related behavior, whether at diabetes onset according to the prediabetes catego- baseline or over follow-up, barely affected the pro- ries. The overall probabilities of diabetes non-onset gression to diabetes onset or reversion to normogly- versus onset were 80.7% vs 19.3%, and the overall cemia according to prediabetes categories odds were 4:1. The highest odds against diabetes (Supplementary tables 2 and 3). onset were observed in individuals with isolated FPG 100-109 mg/dL (29:1) and isolated HbA1c 5.7–5.9% Discussion (20:1). Their probabilities of diabetes non-onset versus onset were, for the isolated FPG 100–109 mg/dL Main findings group, 96.7% (95% CI 93.7–99.7) vs 3.3% (95% CI The evolution of participants with prediabetes over 0.3–6.3); and for the isolated HbA1c 5.7–5.9% group, five-years follow-up was heterogeneous, depending on 95.2% (95% CI 92.3–98.2) vs 4.8% (95% CI 1.8–7.7). The baseline FPG and HbA1c levels. The probability of dia- lowest odds against diabetes onset were observed in betes non-onset/onset ranged from 97%/3%, in people the FPG 110–125 mg/dL plus HbA1c 6.0–6.4% group categorized as prediabetic based on the sole criterion (1:1), whose probabilities of diabetes non-onset versus of isolated FPG levels of 100–109 mg/d-L, to 52%/48% onset were 51.6% (95% CI 44.8–59.5) vs 48.4% (95% in those whose FPG values were 110-125 mg/dL and CI 48.5–55.2). whose HbA1c levels were 6.0–6.4%. As a result, the Table 2 also shows the probability (in percent- odds against diabetes onset were 29:1 in the former age) of reversion to normoglycemia. The overall group and 1:1 in the latter, while the probability of probability was 19.7% (95% CI 17.3-22.0). The high- reverting to normoglycemia was 31.2% and 6%, est probabilities were among those with isolated respectively. HbA1c 5.7–5.9% (34.5%, 95% CI 28.2–40.5) and iso- lated FPG 100–109 mg/dL (31.2%, 95% CI 24.0–39.6), while the lowest was in the category with FPG Strengths and limitations 110–125 mg/dL plus HbA1c 6.0–6.4% (6.2%, 95% The long period of follow-up and the annual assess- CI 1.4–10.0). ments of FPG and HbA1c are strengths of this study. For each category of prediabetes, Figure 1 shows In contrast to other studies reporting diabetes inci- the odds against diabetes onset (converted in prob- dence only as a rate (making it difficult for clinicians ability) as well as the number of participants whose to communicate the prognosis to prediabetic blood glucose and/or HbA1c values remained at pre- patients), in the present study, the prognosis was cal- diabetic levels, normalized, or progressed to diabetic culated according to different prediabetes categories SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 359 Figure 1. Pictograms to explain the prognosis of prediabetes per 100 people at five years’ follow-up in each prediabetes category in the PREDAPS study. Probability of diabetes non-onset versus onset, and number of people who revert to normoglycemia, con- tinue with prediabetes , and who develop diabetes . as a probability. Also the probability of diabetes has lifestyles and health problems at baseline, shown in been compared with the probability of diabetes non- Supplementary table 1, were similar in the subjects onset, which is of great interest to physicians and who completed the study than in those who aban- patients when deciding the alternatives of action. doned it throughout the follow-up. Likewise, the inclusion of probability of reversion to The present study shows the prognosis in subjects normoglycemia is very useful to weigh deci- with prediabetes aged 35 to 74 years. The findings sion-making. could have been different if the lower age limit had In most studies, there are certain overlaps in the been higher. However, no modification was observed criteria used to define prediabetes categories. In this in the relationship between the different categories of study, the prognosis was estimated at different FPG prediabetes analyzed and the prognostic results, after and HbA1c thresholds, using mutually exclusive and performing separate analyzes in subjects under 50 and collectively exhaustive categories. PREDAPS did not in subjects 50 years of age and older. include participants with prediabetes diagnosed based This observational study was carried out in routine on oral glucose tolerance test (OGTT) [19,22], therefore clinical practice. Therefore, a complete absence of it was not possible to estimate the prognosis for dif- intervention cannot be ruled out, as physicians ferent categories according to this criterion. Although advised patients on lifestyle changes based on clinical OGTT values even in isolated form have been shown and laboratory parameters. In any case, the impact on to predict diabetes, it is not a test performed routinely the development of diabetes was small since the inci- in clinical practice. Therefore, its relevance is more for dence rate hardly varied throughout the five years of scientific practice than for routine clinical practice. follow-up. Its magnitude in the first and last year was The results could be biased patients who aban- 4.1 and 4.2 per 100 person-years, respectively. doned were different from those who remained. In subjects with prediabetes, other factors, apart However, the distribution and frequency of age, sex, from FPG and/or HbA1c levels, are relevant for clinical 360 E. REGIDOR ET AL. decision-making. Like social context, housing stability, ADA-established thresholds of 100 mg/dL for glycemia and financial barriers. Such factors have not been con- and 5.7% for HbA1c for defining prediabetes favors sidered in the present study. Physicians, especially in overdiagnosis and confers an unnecessary burden on primary health care, must weigh the knowledge of the health system, as it converts healthy people into these factors in each patient with the information on patients [16–18]. On the other hand, other authors the different results of the evolution of prediabetes argue that these values increase the diabetes onset shown here. risk, and they consider that physicians should advise The present study did not consider other clinical weight loss and physical activity, with or without disorders and health conditions, as for example cardio- medication, to all people with abnormally high indica- vascular events, for which people with prediabetes tors in order to delay or prevent the development of could be at risk, and which should be communicated diabetes [13–14]. Some authors consider it necessary to patients. In the PREDAPS study, the probability of to recommend behavioral modifications even in peo- experiencing any cardiovascular event was 4.9% in the ple with FPG or HbA1c values below these cut- participants with prediabetes, suggesting that the inci- offs [13,14,26]. dence in the lower-risk categories was modest [23]. The ADA thresholds for defining prediabetes were Finally, the analyses did not take into account the established as the result of an intersubjective consen- use of antihypertensive or lipid lowering drugs. The sus among numerous actors from different institutions frequency of use of these drugs was high given the and professional groups. Controversy around the exact high prevalence of hypertension and dyslipidemia in values is likely to continue given the heterogeneity in these subjects. However, the inclusion of these condi- the prognosis for prediabetes in different studies [27] tions in the analysis did not modify the relationship as well as the different criteria applied in clinical prac- between the different categories of prediabetes and tice [28]. Some physicians consider that the the prognostic results (supplementary tables 2 and 3). Hippocratic Oath requires intervention to attend peo- ple who are sick, whereas others interpret it to include care for healthy people, even if most will not develop Implications for clinical practice the disease. In light of this debate, shared decision- The proportion of people with prediabetes who making between people with prediabetes and their develop diabetes or return to normoglycemic levels physicians could be the appropriate option for the varies in the literature [11]. The diverse criteria used to best clinical practice, as part of win-win philosophy. define prediabetes and the different periods of follow- For example, doctors who recommend intervention up explain some of this heterogeneity, while the vari- in people with an FPG of 100–109 mg/dL and/or HbA1c able distribution of prediabetes phenotypes from one of 5.7–5.9% consider that these people will benefit place to another also need to be considered. For from preventive care. According to one study, behav- example, about 20% of people with prediabetes in the ioral modification in such populations can reduce the Japanese population presented altered levels of both relative risk of developing diabetes to a similar extent FPG and HbA1c [10,24], compared to 50% in the as in people with higher values [29]. However, consider- PREDAPS study [19,22], a difference that could explain ing that the implementation of these measures is bene- the higher five-year incidence of diabetes in the ficial implies a value judgement. Spanish compared to the Japanese cohort. However, According to the findings found here, 97% of peo- the prognosis was similar in both populations when ple diagnosed with prediabetes based solely on FPG analyzed according to prediabetes phenotypes values, and 95% diagnosed based on HbA1c, will not [10,19,24], suggesting that our results are probably develop diabetes within five years. Most of them quite similar to other geographical settings. would not consider it beneficial to attempt to increase In observational studies following people with pre- their chances of not developing diabetes through diabetes for 1 to 10 years, most do not develop dia- behavioral modifications. This is despite the fact that betes, and a large proportion returned to reversion to normoglycemia could increase from normoglycemia [9,11]. In the present study, the prob- 31–35% to 45%, in line with results from a meta-ana- ability of diabetes non-onset was 80.7%, while the lysis of intervention studies that combined changes in probability of reverting to normoglycemia was 19.7%. diet and physical activity, yielding a relative risk of Other studies with similar follow-up periods have reversion of 1.53 [30]. Something different happens in reported similar results [10,11,24,25]. These results people who belong to other categories of prediabetes. have prompted some authors to consider that the For example, 52% of the people with FPG of SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 361 110–125 mg/dL plus HbA1c of 6.0–6.4% will not In summary, judgements and values associated develop diabetes. Considering the relative risk of dia- with interpreting findings from scientific knowledge betes incidence of 0.59 obtained in the previously about prediabetes and its application to clinical mentioned meta-analysis [30], this percentage would practice vary among different physicians. Moreover, increase to 71% with behavioral modifications. Thus, these value judgements may be different from their these people would likely consider this potential patients’ preferences. Communicating information on increase to be of great value. the prognosis of prediabetes in different ways that Authors have also taken a public health perspective are understandable could be useful for assisting to support the appropriateness of interventions in people with prediabetes and their physicians make low-risk individuals, as this contributes to reducing informed and shared decisions about different inter- diabetes incidence in the population [31]. However, it vention options. is unreasonable to attribute efficiency to the option of dedicating societal resources to ensure adherence to PREDAPS study group healthy behaviors in people whose risk of developing  Margarita Alonso (CS De la Eria, Asturias), Beatriz Alvarez (CS diabetes is less than 5%. Although some studies justify  Andres Mellado, Madrid), Fernando Alvarez (CS La Calzada 2, the ADA cutoffs as cost-effective, this research has not  Asturias), J Carlos Alvarez (CS Eras de Renueva, Leo n), Ma del specifically looked at the cost-effectiveness of inter-  Mar Alvarez (CS Hereza, Madrid), J Joaquın Anton (CS Murcia- ventions in the prediabetes categories associated with Centro, Murcia), Oriol Armengol (EAP Poblenou, Barcelona), a lower risk of developing the disease [32–34].  Sara Artola (CS Jose Marva, Madrid), Luis Avila (Consultorio Implementing preventive interventions in healthy Almachar, Malaga), Carmen Babace (CS Rodrıguez Paterna, La people –with doubtful clinical justification- is quite dif- Rioja), Lourdes Barutell (CS Andres Mellado, Madrid), Ma Jesus Bedoya (CS Hereza, Madrid), Belen Benito (EAP Raval ferent from treating a patient who has sought help for Sud, Barcelona), Beatriz Bilbeny (EAP Raval Sud, Barcelona), a specific problem [35]. This task requires physicians Marti Birules (EAP Poblenou, Barcelona), Concepcio n Blanco to improve their understanding and ability to commu- (CS Sada, A Corun ~a), Ma Isabel Bobe (EAP La Mina, nicate about the risk of the disease they seek to pre- Barcelona), Carmen Boente (CS Porrin ~o, Pontevedra), Antonia vent and the absolute risk reduction that the possible Borras (CS Canal Salat, Baleares), Remei Bosch (EAP Girona 2, Girona), Ma Jesu s Brito (CS La Matanza, Baleares), Pilar Buil intervention entails [36]. Physicians also need (EAP Azpilagan ~a, Navarra), J Jose Cabre (EAP Reus-1, adequate ways to present the information to ensure Tarragona), Ainoha Cambra (CS Arrabal, Zaragoza), Francisco its usefulness during clinical decision-making. Carbonell (CS Mislata, Valencia), Francisco Carramin ~ana (CS The first step should be to transmit information San Roque, Badajoz), Lourdes Carrillo (CS La Victoria de about the natural history of the disease. The number Acentejo, Santa Cruz de Tenerife), Ana Casorran (CS Fuente de San Luis, Valencia), Rafael Colas (CS Santon ~a, Cantabria), absolutes shown here in pictograms constitute a sim- Blanca Cordero (CS Sta. Marıa de Benquerencia, Toledo), ple way for people with diverse levels of prediabetes Xavier Cos (EAP Sant Martı de Provençals, Barcelona), Gabriel indicators to understand their chances of developing Cuatrecasas (CAP de Sarria, Barcelona), Cristina De Castro (CS diabetes, returning to normal glycemic levels, or con- Sta. Marıa de Benquerencia, Toledo), Manuel De la Flor (CS tinuing in the same prediabetic situation. Moreover, Ntra. Sra. de Gracia, Sevilla), Carlos De la Sen (Consultorio the information about the probabilities against dia- San Gabriel, Alicante), Rosa Mar De Miguel (EAP Pubillas Casas, Barcelona), A Marıa De Santiago (Unidad Docente de betes onset can provide valuable information for sub- Atencio n Familiar y Comunitaria, Guadalajara), Mercedes Del jective judgements about whether or not to adopt the Castillo (CS Andres Mellado, Madrid), Ma Carmen Duran (CS preventive measures that their doctors proposals. On Lavadores Vigo, Pontevedra), Patxi Ezkurra (CS Zumaia, hearing this information, some people with prediabe- Guipu zcoa), Paula Gabriel (EAP Badia del Valles, Barcelona), tes may not even want to know about the different Javier Gamarra (CS Medina del Campo Rural, Valladolid), options their physicians describes. Francisco Garcıa (CS Don Benito Este, Badajoz), Luis Garcıa- Giralda (CS Murcia Centro, Murcia), F Javier Garcıa-Soidan (CS A second stage would provide physicians with the ~o, Pontevedra), Ma Teresa Gijo n (CS Los Yebenes, Porrin existing evidence supporting the intervention(s) that Madrid), Albert Goday (Endocrinologıa Hospital del Mar, they recommend. In this case, they have no choice  Barcelona), Angel mez (CS Lasarte, Guipu Go zcoa), Marıa del but to try to make these recommendations compatible Carmen Go mez (CS Velez-Malaga Norte, Malaga), J Carles with the odds of probabilities and the absolute num- Gonzalez (Girona 3, Girona), Marıa Gonzalez (CS Alcantarilla Sangonera, Murcia), Esteban Granero (CS Vista Alegre Murcia, bers. However, most intervention studies about behav-  Murcia), Angela Trinidad Gutierrez (CS El Calero, Las Palmas), ioral changes and/or drug prescriptions do not Felix Gutierrez (CS Bombarda-Monsalud, Zaragoza), Luisa disaggregate the results according to various catego- Gutierrez (CS Beraun, Guip  uzcoa), M Angel  Gutierrez (CS Avila ries of prediabetes.  Sur Oeste, Avila), Ana M Hernandez (CS El Calero, Las a 362 E. REGIDOR ET AL. Palmas) , Mercedes Iban ~ez (CS Vandel, Madrid), Rosario Author contributions Iglesias (CS Lain Entralgo, Madrid), Dimas Igual (CS Manuel Encinas, Caceres), Jaime Innenaraty (CS Hereza Leganes, ER and LCS analyzed and interpreted the data. LCS, ER and Madrid), Jaime Iriarte (CS Aizarnazabal-Getaria, Guipu zcua), JFN wrote the outline and edited the manuscript. AR, AG,  Angeles Jurado (CS Salvador Caballero, Granada), Rafael DC, JDE and SA reviewed/edited the manuscript and contrib- Llanes (Villanueva de la Can ~ada, Madrid), Flora Lo pez (EAP uted to discussion. As reviewed and edited the manuscript. Martorell, Barcelona), Regina Lo pez (EAP El Carmel, All authors read and approved the final manuscript Barcelona), Riansares Lo  pez (CS Artilleros, Madrid), Angela Lorenzo (CS Alcala de Guadaira, Madrid), Carmen Losada n Macıa (CS Roces Montevil, Disclosure statement (UGC Adoratrices, Huelva), Ramo Asturias), Fernando Malo (CS Ares, A Corun ~a), Jose Mancera The declare that there is no conflict of interest that could be (CS Ciudad Jardın, Malaga), Ma Jose Mansilla (CS Martın de perceived as prejudicing the impartiality of the Vargas, Madrid), Ma Teresa Marın (CS General Ricardos, research reported. Madrid), Jose Luis Martın (CS Salvador Caballero, Granada), Ana Martınez (EAP El Carmel, Barcelona), F Javier Martınez (CS Federica Monseny, Madrid), Juan Martınez (CS Yecla, Funding Murcia), Ma del Carmen Martınez (EAP Raval Sud, Barcelona), Sanofi and Novartis gave financial support for the develop- Rosario Martınez (CS On ~ati, Guip uzcoa), Anna Massana (EAP ment of the data collection platform, the meetings of Raval Sud), Ma Soledad Mayayo (CS Martın de Vargas, researchers, and the monitoring of information collected at Madrid), J Javier Mediavilla (CS Burgos Rural, Burgos), Luis baseline. Sanofi and Novartis had no role in the study Mendo (CS Cadreita, Navarra), J Manuel Millaruelo (CS design, analysis and interpretation of data, writing of the Torrero La Paz, Zaragoza), Alicia Monzo  n (CS Vecindario, Las manuscript, nor the decision to submit the manuscript for Palmas) , Ana Moreno (CAP San Roque, Badajoz), Pedro publication. Mun ~oz (Unidad Docente de Medicina Familiar y Comunitaria, Cantabria), Xavier Mundet (EAP Carmel, Barcelona), Teresa Mur (CAP Terrasa Sud, Barcelona), Emma Navarro (CS An ~aza, Santa Cruz de Tenerife), Jorge Navarro (CS Salvador Pau, References  Valencia), Pedro Nogales (CS Las Aguilas, Madrid), J Carlos Obaya (CS Chopera, Madrid), Cristina Oria (CS Aizarnazabal- [1] Herman CR, Gill HK, Eng J, et al. Screening for preclin- Guetaria, Guip uzcoa), Francisco Javier Ortega (CS Campos- ical disease: test and disease characteristics. AJR Am J Lampreana, Zamora), Francisca Paniagua (CS Ciudad Jardın, Roentgenol. 2002;179(4):825–831. Malaga), Jose Luis Pardo (CS Orihuela I, Alicante), Francisco [2] Herman C. What makes a screening exam "good"? Carlos Perez (CS Martın de Vargas, Madrid), Pedro P Perez (CS Virtual Mentor. 2006;8(1):34–37. Mallen, Sevilla), Neus Piulats (EAP Raval Sud, Barcelona), [3] Fagerlin A, Zikmund-Fisher BJ, Ubel PA. Helping Raquel Plana (CS Ponteareas, Pontevedra), Nuria Porta (CAP patients decide: ten steps to better risk communica- Terrassa Sud, Barcelona), Santiago Poveda (CS Jumilla, tion. J Natl Cancer Inst. 2011;103(19):1436–1443. Murcia), Luis Prieto (CS Caceres-La Mejostilla, Caceres), [4] Mostofsky E, Dunn JA, Hernandez-Dıaz S, et al. Patient Ramo n Pujol (EAP Tremp, Lleida), Sol Reixa (CS Arrabal, and Physician Preferences for Reporting Research Zaragoza), Jazmın Ripoll (CS Fuente de San Luis, Valencia), Findings. Fam Med. 2019;51(6):502–508. Antonio Rodrıguez (EAP Angles, Girona), J Jose Rodrıguez (CS [5] Hembroff LA, Holmes-Rovner M, Wills CE. Treatment Villaviciosa de Odo n, Madrid), Ma Angeles Rollan (CS Los decision-making and the form of risk communication: Yebenes, Madrid), Ma Teresa Rollan (CS Hereza, Madrid), results of a factorial survey. BMC Med Inform Decis Laura Romera (EAP Raval Nord, Barcelona), Pilar Roura (EAP Mak. 2004;4:2016 Badia de Valles, Barcelona), Jo se Felix Rubio (CS Lasarte, [6] Trevena LJ, Davey HM, Barratt A, et al. A systematic Guip uzcua), Irene Ruiz (EAP La Torrassa, Barcelona), Manuel review on communicating with patients about evi- Antonio Ruiz (CS Agost, Alicante), Isabel Saenz (CS dence. J Eval Clin Pract. 2006;12(1):13–23. Espronceda, Madrid), Julio Sagredo (CS Los Rosales, Madrid), [7] Goodyear-Smith F, Kenealy T, Wells S, et al. Patients’ Alejandro Salanova (CS Fuente de San Luis, Valencia), L preferences for ways to communicate benefits of car- Gabriel Sanchez (CS Carballeda, Zamora), Manuel Sanchez diovascular medication. Ann Fam Med. 2011;9(2): (CS Vista Alegre Murcia, Murcia), Gloria Sanz (CS San Jose 121–127. centro, Zaragoza), Mateu Seguı (UBS Es Castell, Baleares), [8] American Diabetes Association. Standards of medical Josefina Servent (EAP El Carmel, Barcelona), Dulce Suarez (CS care in diabetes – 2020. Diabetes Care. 2020; El Calero, Las Palmas) , Eduard Tarrago  (EAP Bellvitge, 43(Suppl 1):S14–S31. Barcelona), Jesu s Torrecilla (CS Bombarda-Monsalud, [9] Tabak AG, Herder C, Rathmann W, et al. Prediabetes: Zaragoza), Jose Luis Torres (CS Rodrıguez Paterna, La Rioja), a high-risk state for diabetes development. Lancet. Mere Villaro (CAP Terrassa sud, Barcelona), Carmen Yuste 2012;379(9833):2279–2290. (EAP Badia del Valles, Barcelona). [10] Heianza Y, Hara S, Arase Y, et al. HbA1c 5.7-6.4% and Guarantor’s: E. Regidor is the guarantor of this work and, impaired fasting plasma glucose for diagnosis of pre- as such, had full access to all the data in the study and takes diabetes and risk of progression to diabetes in Japan responsibility for the integrity of the data and the accuracy (TOPICS 3): a longitudinal cohort study. Lancet. 2011; of the data analysis. 378(9786):147–155. SCANDINAVIAN JOURNAL OF PRIMARY HEALTH CARE 363 [11] Richter B, Hemmingsen B, Metzendorf M-I, et al. cut-off points for HbA(1c) and impaired fasting glu- Development of type 2 diabetes mellitus in people cose: the Toranomon Hospital Health Management with intermediate hyperglycaemia. Cochrane Center Study 4 (TOPICS 4)). Diabet Med. 2012;29(9): Database Syst Rev. 2018;10:CD012661. e279–e285. [12] Shang Y, Marseglia A, Fratiglioni L, et al. Natural his- [25] Brunisholz KD, Joy EA, Hashibe M, et al. Incidental risk tory of prediabetes in older adults from a population- of type 2 diabetes mellitus among patients with con- based longitudinal study. J Intern Med. 2019;286(3): firmed and unconfirmed prediabetes. PLoS One. 2016; 326–340. 11(7):e0157729. [13] Arky RA. "Doctor, is my sugar normal?" N Engl J Med. [26] Tuso P. Prediabetes and lifestyle modification: time to 2005;353(14):1511–1513. prevent a preventable disease. Perm J. 2014;18(3): [14] Bergman M. Inadequacies of absolute threshold levels 88–93. Summer for diagnosing prediabetes. Diabetes Metab Res Rev. [27] Zhang X, Gregg EW, Williamson DF, et al. A1C level 2010;26(1):3–6. and future risk of diabetes: a systematic review. [15] Cefalu WT. "Prediabetes": Are There Problems With Diabetes Care. 2010;33(7):1665–1673. This Label? No, We Need Heightened Awareness of [28] Hostalek U. Global epidemiology of prediabetes - pre- This Condition!!. Diabetes Care. 2016;39(8):1472–1477. sent and future perspectives. Clin Diabetes [16] Yudkin JS, Montori VM. The epidemic of pre-diabetes: Endocrinol. 2019;5:5. the medicine and the politics. BMJ. 2014;349:g4485. [29] Knowler WC, Barrett-Connor E, Fowler SE, Diabetes [17] Van den Bruel A. The triumph of medicine: how over- Prevention Program Research Group, et al. Reduction diagnosis is turning healthy people into patients. Fam in the incidence of type 2 diabetes with lifestyle inter- Pract. 2015;32(2):127–128. vention or metformin. N Engl J Med. 2002;346(6): [18] Yudkin JS. "Prediabetes": Are There Problems With 393–403. ;. This Label? Yes, the Label Creates Further Problems! [30] Balk EM, Earley A, Raman G, et al. Combined Diet and Diabetes Care. 2016;39(8):1468–1471. Physical Activity Promotion Programs to Prevent Type [19] Serrano R, Garcıa-Soidan FJ, Dıaz-Redondo A, et al.; 2 Diabetes Among Persons at Increased Risk: A Grupo de Estudio PREDADS. [Cohort Study in Primary Systematic Review for the Community Preventive Health Care on the Evolution of Patients with Services Task Force. Ann Intern Med. 2015;163(6): Prediabetes (PREDAPS): basis and methodology]. Rev Esp Salud Publica. 2013;87(2):121–135. 437–451. [20] Breslow NE, Day NE. Statistical methods in cancer [31] Waugh N. Screen and intervene to prevent diabetes? research, volume II. IARC Sci Publ. 1987;82:1–406. Bmj. 2017;356:i680. [21] Kleimbaum DG, Kupper LL, Morgenstern H. [32] Zhuo X, Zhang P, Selvin E, et al. Alternative HbA1c Epidemiology research. Principles and quantitative cutoffs to identify high-risk adults for diabetes pre- methods. New York: Van Nostrand Reinhold vention: a cost-effectiveness perspective. Am J Prev Company, 1982. 96–116. Med. 2012;42(4):374–381. [22] Giraldez-Garcıa C, Sangro s FJ, Dıaz-Redondo A, et al.; [33] Zhuo X, Zhang P, Kahn HS, et al. Cost-effectiveness of PREDAPS Study Group. Cardiometabolic Risk Profiles alternative thresholds of the fasting plasma glucose in Patients With Impaired Fasting Glucose and/or test to identify the target population for type 2 dia- Hemoglobin A1c 5.7% to 6.4%: Evidence for a betes prevention in adults aged 45 years. Diabetes Gradient According to Diagnostic Criteria: The Care. 2013;36(12):3992–3998. PREDAPS Study. Medicine (Baltimore). 2015;94(44): [34] Glechner A, Keuchel L, Affengruber L, et al. Gartlehner e1935. ;. GEffects of lifestyle changes on adults with prediabe- [23] Giraldez-Garcıa C, Hernandez AM, Gamarra J, et al. tes: A systematic review and meta-analysis. Prim Care Evolucion de pacientes con prediabetes en Atencio n Diabetes. 2018;12(5):393–408. Primaria de Salud (PREDAPS): resultados del quinto [35] Martyn C. Risky business: doctors’ understanding of ~o de seguimiento. Diabetes Practica. 2018;09:37–80. an statistics. BMJ. 2014;349:g5619. [24] Heianza Y, Arase Y, Fujihara K, et al. Screening for [36] J€arvinen TL. Labelling people as ’High Risk’: A tyranny pre-diabetes to predict future diabetes using various of eminence? Br J Sports Med. 2016;50(2):77–78.