Polycystic ovary syndrome (PCOS) is a common female endocrinopathy, recognised as a heterogeneous disorder, characterised by hyperandrogenism, ovulatory dysfunction and polycystic ovary ultrasound image(Reference Sirmans and Pate1). The prevalence of PCOS ranges between 8 and 13 % around the world and indicates that it has become a major public health issue worldwide(Reference Bozdag, Mumusoglu and Zengin2). About 50 % of diagnosed women are obese or overweight(Reference Azziz, Woods and Reyna3). Consequently, PCOS women are at increased risk of insulin resistance and metabolic syndrome(Reference Pasquali4).
The choice of treatment is related to the symptoms such as menstruation abnormalities, androgen excess and infertility(Reference Sirmans and Pate1,Reference Knochenhauer, Key and Kahsar-Miller5) . One of the main goals of PCOS therapy is to decrease metabolic consequences related to obesity, fatness, insulin resistance and metabolic syndrome. Therefore, it is important to reduce excess of fat mass related to obesity and insulin sensitivity in PCOS women. According to the Androgen Excess Society and the European Society of Endocrinology, PCOS therapy should be based on lifestyle changes, including increased physical activity and dietary modifications concerning the intake of SFA, low Glycaemic Index (GI) foods and weight reduction if necessary(Reference Conway, Dewailly and Diamanti-Kandarakis6,Reference Salley, Wickham and Cheang7) . Following this, diet quality recommendations and lifestyle modifications for PCOS women should be the prior treatment approach. Despite many studies, the optimum and effective diet composition has not yet been identified. A limited number of papers analysing the correlation between diet quality and body composition in PCOS women are available(Reference Pasquali, Gambineri and Biscotti8,Reference Cunha, da, Ribeiro and Silva9) . The only confirmed beneficiary strategy for obese PCOS women was concluded by Moran et al. (Reference Moran, Ko and Misso10) and indicated weight loss, without defined diet composition. Other studies suggest instead the role of the particular features of the diet like GI, energy intake reduction, protein and carbohydrates in the management of PCOS(Reference Graff, Mario and Magalhães11–Reference Milewicz, Kudła and Spaczyński18). Only a few studies analysed the dietary patterns of PCOS women, while their relation to body fatness (BF) had not been studied yet(Reference Shahdadian, Ghiasvand and Abbasi19,Reference Huijgen, Louwers and Willemsen20) . Bearing in mind the fact that diet is the composition of many nutrients, it is increasingly recognised that whole dietary scores are a more important predictor of non-communicable disease risk(Reference Reedy, Krebs-Smith and Miller21–Reference Milajerdi, Namazi and Larijani23). Moreover, the analysis of overall dietary scores more closely parallels real life, where subjects do not take isolated nutrients but rather as meals consisting of a variety of foods with complex combinations of nutrients which may be interactive or synergistic. The current approach to examine the link between nutrition and diseases recommends moving from single nutrients to dietary patterns, including a group of foods as diet quality scores (DQS). For this reason, the nutritional management of PCOS should be discussed and debated under consideration of overall diet quality and then selected nutrients and their relation to the BF.
Dietary patterns are influenced by social, physiological and economic factors and vary between individuals and populations(Reference Birkenhead and Slater24,Reference Kabir, Miah and Islam25) . The status and critical role of nutrition knowledge (NK) have been hypothesised and are still uncertain in public health nutrition. Health abnormalities might be an additional significant factor influencing their dietary choices. As proved in previous studies, nutrition education can cause the change in nutrition behaviours(Reference Hamulka, Wadolowska and Hoffmann26), body weight and prevalence of obesity(Reference Valmorbida, Goulart and Busnello27,Reference Bonaccio, Di Castelnuovo and Costanzo28) . To the best of our knowledge, no studies were investigating whether the NK of PCOS women could be related to their dietary behaviours. Considering that the search of mutual relations between DQS, BF and NK will give a possibility to plan, complementary therapy addresses metabolic abnormalities of PCOS women.
The limited amount of literature in this area provides a basis for further research on the recommended composition of the diet in relation to BF and NK of PCOS women. Therefore, based on the presented studies, it is essential to determine the relation of the DQS, NK and BF in PCOS and compare these with a group of healthy women with similar age and socio-economic status.
Materials and methods
Participants
This case–control study included 122 women with PCOS (PCOS) and 116 age and socio-economically matched healthy controls (CON). Participants were selected and diagnosed in the Department of Endocrinology, Metabolism and Internal Diseases. The patients flow chart is shown in Fig. 1. Women with PCOS were classified according to the Rotterdam criteria(29), which are the most common diagnostic criteria for PCOS. Rotterdam criteria recommend PCOS to be diagnosed when at least two of the following three features were present: (1) oligoovulation; (2) hyperandrogenism (elevated androgen levels: testosterone >2·67 nmol/l, free testosterone >11 pmol/l and/or free androgen index >5·5) or hyperandrogenisation (acne and hirsutism) and/or (3) ultrasonographic findings: the presence of at least twelve follicles in each ovary measuring 2–9 mm in diameter, and/or ovarian volume more than 10 ml. Lack of food allergies and intolerances were inclusion criteria. The CON women were a convenient sample of female volunteers recruited through advertising at the university, local workplaces and public spaces. Inclusion criteria for CON were normal ovulating cycles with no signs of hyperandrogenism. Chronic hepatic, renal and rheumatic diseases, overt hypothyroidism, steroid therapy in the last 3 months and pregnancy were exclusion criteria for both PCOS and CON groups. The use of oral contraceptives was also among exclusion criteria for both groups, due to the inability of PCOS diagnosis while using this type of pharmacotherapy. The same day after endocrinologist appointment, all the participants were directed to the Institute of Human Nutrition and Dietetics, Poznan University of Life Sciences, where the body composition was measured and a dietary interview was performed. The study appointments took place in the mornings, between October 2016 and January 2018.
Fig. 1. Study recruitment process
One hundred seventy-five patients were screened towards PCOS diagnosis between 2016 and 2018, from which 134 patients had positive PCOS diagnosis. One hundred forty-six subjects were self-reported as healthy, but after the screening process, six participants were rejected. During the matching process (described in the Statistics part), twelve participants with PCOS and twenty-four healthy participants were rejected due to age and/or socio-economic status. The BMI of the rejected participants was no different than the study group.
The diet quality scores
The DQS were calculated as dietary indexes, collected with a validated questionnaire (the Dietary Habits and Nutrition Beliefs KomPAN Questionnaire) developed by the Committee of Human Nutrition, Polish Academy of Science(Reference Jeżewska-Zychowicz, Gawęcki, Wądołowska and Gawecki30). The interview was conducted by a qualified pollster. Closed questions regarding the frequency of consumption of thirty-three food groups were used in the questionnaire. Six food frequency intake categories were used: (1) never, (2) 1–3 times/month, (3) once a week, (4) 2–3 times/week, (5) once a day and (6) a few times during the day. For each food item, the categories of frequency consumption were converted to values reflecting daily frequency consumption (1) never = 0·00, (2) 1–2 times/month = 0·06, (3) once a week = 0·14, (4) 2–3 times/week = 0·5, (5) once a day = 1 and (6) a few times during the day = 2(Reference Wadolowska, Ulewicz and Sobas31,Reference Wadolowska, Hamulka and Kowalkowska32) . Conversion rates have been arbitrarily accepted by literature analysis and the authors’ own experience(Reference Kowalkowska, Wadolowska and Czarnocinska33). The description of food frequency intake was undertaken by using the diet indexes characterising the intensity of food frequency intake of chosen groups of products widely known as healthy, unhealthy, with a high- and low-GI, high content of simple sugars and saturated fats. To evaluate overall diet quality, a pro-Healthy-Diet-Index (pHDI-10) and a non-Healthy-Diet-Index (nHDI-14) scores were established based on previous knowledge and other validated studies(Reference Gawęcki34). Due to multiple reports on its effects in PCOS management, high- and low-GI foods as well as foods with high saturated fats and high simple sugar were analysed(Reference Moran, Ko and Misso10,Reference Graff, Mario and Magalhães11,Reference Graff, Mário and Alves35) . To elucidate dietary intake of high- and low-GI foods, high-sugar, and high saturated fat dietary sources, the four novum DQS were created by the authors based on the previously validated scores (pHDI-10, n-HDI-14) (Table 2). Each index was designed and calculated based on the same conversion formula as previously validated(Reference Kowalkowska, Wadolowska and Czarnocinska33):
$${{{Diet\;Quality\;Index}} = {\;{{100\% *\sum A}}\over{{\sum \;B}}}\;\left[ \% \right]$$
where A is the sum of the reported daily intake of all items listed in specific food groups (e.g. low GI, see Table 2), for example, ∑ = 0 + 0·14 + 0·06 + 0·5. B is the sum of the maximum possible to report daily intake of the same (low GI) foods, determined for one product as 2 (e.g. ∑ = 2 + 2 + 2 + 2).
All dietary scores were created a priori by summing the consumption frequencies (times/d) of the following food items: the pHDI – dairy products, fish, vegetables, fruit; the nHDI – fast food, sweetened soft drinks, energy drinks, and sweets, both already validated(Reference Kowalkowska, Wadolowska and Czarnocinska33). For this study, the high-Glycaemic-Diet-Index-7 (hGIDI-7), low-Glycaemic-Diet-Index-4 (LGIDI-4), high-Sugar-Diet-Index-4 (hSDI-4) and the high-saturated-fats-Diet-Index-8 (hSFDI-8) indexes were designed. Groups of food products classified in individual diet indexes are presented in Table 2. The consumption levels of the food groups (indexes) were classified on a percentage scale. Each DQS was expressed in % points and was categorised as follows: low (0–33·32 % points), moderate (33·33–66·65 % points) and high (66·66–100 % points) intensity of consumption of selected food groups. For example, the calculation for low-Glycaemic-Diet-Index-4 (LGIDI-4) was: lGIDI-4 = 100 %*0·7/8 = 8·75 %. There was a low intensity of consumption of low GI foods in the studied group.
Body fatness
The subject’s BF was assessed by the air displacement plethysmography method using BodPod (Life Measurement Inc.). During the measurement, all the subjects wore swimming suits and swimming caps; they fasted for 2 h before the analysis(Reference Biaggi, Vollman and Nies36). The measurements were taken in morning hours. The reference values for the body fat (BF) percentage were accepted from the literature(Reference Jeukendrup and Michael37) in the three ranges: normal (below 30 %), above normal (30–36 %) and high risk (above 36 %). The basic anthropometric measurements were also taken: body mass (kg), height (cm), waist and hip circumference (cm).
Nutrition knowledge
The nutrition beliefs and knowledge of the studied groups were determined by using the validated KomPAN questionnaire(Reference Kowalkowska, Wadolowska and Czarnocinska33). It included the twenty-five statements and assessments expressed as ‘truth’, ‘false’ and ‘hard to say’. For each correct answer, the respondent received 1 point, and for an incorrect answer or ‘hard to say’, 0 point. Points were summed up for each respondent (range 0 to 25 points). The respondents were divided into three nutrition knowledge categories: insufficient (0–8 points), sufficient (9–16 points) and good (17–25 points).
Physical activity
To assess physical activity, the short version of the international physical activity questionnaire was used(Reference Lee, Macfarlane and Lam38).
Statistics
To calculate the sample size, the Sample Size Calculator Clinical Calc (ClinCalc, LLC) was used(Reference Sean39). The anticipated means of the pHDI-10 score were set as 27 ± 15 (low score) and 34 (high score). The minimum number of subjects for adequate study power was calculated as 119 for each independent group with the enrolment ratio set at 1, type I error at 0·05 and power 95 %.
To achieve comparability between the study groups (PCOS v. CON), propensity score 1:1 matching was performed, in which patients in the PCOS group were matched with counterparts in the CON group according to age and socio-economic status. The subjects were matched according to the closest distance with the replacement of the matched control subject(Reference Deb, Austin and Tu40). The normality of the distribution data was verified with the Shapiro–Wilk test. The logistic regression was used to determine the OR for DQS and body fat by nutrition knowledge. Differences in nutritional status and frequency intake were calculated using the t test of two independent samples. For non-normal distribution data, the Mann–Whitney U test was used. Values are presented as mean and median. The 95 % CI was set. A P value of <0·05 was considered to be statistically significant. The statistical analysis was carried out using Statistica 13·1 software (Dell Inc. 2016).
Results
As depicted in Table 1, the general characteristics show that most of the women participated in the study lived in cities with more than 20 000 residents, declared an average economic situation, had higher education level and lived in 1–4 person households. In the PCOS group, more women smoked cigarettes. As shown in Table 1, over 40 % of women with PCOS were overweight or obese and presented a higher BF.
Table 1. Sample characteristics between women with polycystic ovary syndrome (PCOS) and control group
* P < 0·05.
† Variables categorised into clases (as percentage and number of subjects in presented category, e.g. age ≤ 24·9).
‡ Results from the χ 2 test.
§ Physical activity was categorised according to the manual of a short version of International Physical Activity Questionnaire (IPAQ-SF)(Reference Lee, Macfarlane and Lam38).
Table 2. Characteristics of dietary indexes
Table 3 shows the significant anthropometric differences between the PCOS and CON groups. The women with PCOS had significantly higher body fat and BMI. The waist:height ratio was significantly higher in comparison with the CON group. Women with PCOS declared less intense physical activity and more sitting time than the control group. Ninety-one percentage of the PCOS group were diagnosed with hyperandrogenism, 76 % had oligoovulation, 91 % irregular menstruation, 13 % hyperlipidaemia and 34 % insulin resistance.
Table 3. Anthropometrics, physical exercise and body composition measurements between polycystic ovary syndrome (PCOS) and control group
WHtR, waist:height ratio.
† Level of significance for the comparison of means between groups: *P < 0·05.
‡ MET-minute/week.
As shown in Table 4, both groups presented low intensity of consumption of foods perceived as generally healthy. The PCOS group had a significantly lower frequency intake of pro-healthy foods (pHDI-10) and low GI foods (LGIDI-4) than the CON group. There was a tendency in a higher frequency of high-sugar product (hSFDI-8) intake in the PCOS group than the CON group. PCOS had a lower intake of fruits, vegetables and cottage cheese, what has been shown in detailed components of the indexes included in the online supplementary material, Supplementary Table 6. To ensure that the frequency of eating is not only related to the higher body mass of PCOS women, we stratified the PCOS and CON groups by BMI: 1. BMI < 25 and BMI ≥ 25 (PCOS) and 2. BMI < 30 and BMI ≥ 30 (CON). After performing a Mann–Whitney U test in the CON group, the PCOS group and combined, we did not find any statistical differences between the DQS stratified by BMI. The subjects from the PCOS group had lower average (sufficient level) NK compared with the CON group (at a good level) according to the validation protocol(Reference Kowalkowska, Wadolowska and Czarnocinska33), as shown in Table 4.
Table 4. Diet indexes and nutritional knowledge in polycystic ovary syndrome (PCOS) and CON women
pHDI-10, Pro-Healthy-Diet-Index-10; nHDI-14, Non-Healthy-Diet-Index-14; hGIDI-7, High-Glycemic-Diet-Index-7; hSDI-4, High-Sugar-Diet-Index-4; hSFDI-8, High-Saturated-Fats-Diet-Index-8; lGIDI-4, Low-Glycemic-Diet-Index-4; NK, nutrition knowledge.
† Level of significance for the comparison of means between groups: *P < 0·05.
CON women with at least a good level of NK were more than two times likely to have a high intensity of intake of pro-healthy foods (pHDI-10) and had less than a half likely to have a high intensity of intake of saturated fat dietary sources (hSFDI-8) (Table 5). There was also a tendency in controls with at least good knowledge to have a high intensity of consumption of low GI foods (lGIDI-4) and low intensity of high GI foods. These results did not repeat in the PCOS group.
Table 5. OR (95 % CI) for diet quality scores and body fat by nutrition knowledge in polycystic ovary syndrome (PCOS) women and healthy control group
Level of significance for the comparison of means between groups: *P < 0·05.
† Body fat content categories according to Jeukendrup and Michael(Reference Jeukendrup and Michael37).
‡ High intensity of frequency intake refers to upper tertile of each of index.
§ Nutrition knowledge referred to as good according to the validated questionnaire manual.
Discussion
Our study revealed the dietary habits identified by DQS in the scope of fatness and NK of PCOS women. To the best of our knowledge, this research is the first study to identify using the a priori approach, the dietary intake in PCOS women through integrated nutrition quality scores covering groups of food items. Based on the frequency intake of thirty-three food groups, DQS were grouped and categorised according to consumed food products: pro-healthy, non-healthy, low- and high-GI, a high simple sugar, high SFA. PCOS women, in comparison with CON subjects, presented lower intensity of consumption of foods recommended as ‘generally healthy’ (pHDI-10, Pro-Healthy-Diet-Index-10) and foods characterised by low GI (lGDI-4, Low-Glycemic-Diet-Index-4). We found no relation between DQS, nutrition knowledge and body fat percentage among PCOS women, which was apparent in the healthy control group. Concurrently, we observed a higher body fat and lower nutrition knowledge in the PCOS group.
The first dietary characteristics in PCOS women we present were the low-intensity consumption of foods perceived as healthy. It was found that fruits, vegetables and cottage cheese intake were the main factors involved in lowering pHDI-10. Other characteristics were the low intensity of consumption of low GI foods, recommended for consumption for PCOS, and included whole-wheat bread, whole-wheat cereals, oatmeal, whole-wheat pasta, dishes with legumes and vegetables. The vegetables were the most significant group to differentiate PCOS and controls. In spite of above food groups, PCOS women declared a higher intensity (30 % more than control) of consumption of fast food. Our results suggest that PCOS women were prone to a less healthy lifestyle; they were smokers more often than CON and engaged in less intense physical activity but were sitting more. Such picture of studied PCOS was supplemented by 20 % and 22 % lower consumption of fruit and vegetables in this group. This tendency was also observed by Shishehgar et al.(Reference Shishehgar, Ramezani Tehrani and Mirmiran41). According to the current nutrition guidelines for the general population, the majority of food groups classified in the healthy and low glycaemic intake index (pHDI-10 and lGIDI-4) should be consumed at least several times a day. Food products classified in the nonhealthy dietary index (nHDI-14) and high GI (hGIDI-7) (white bread, white rice, plain pasta or cereal, fast food, butter, lard, cheese, sausages, red meat, sweets, canned meat, sweetened carbonated and non-carbonated drinks, energy drinks and alcoholic beverages) should be consumed in the smallest possible amounts. Dietary intake profile of our PCOS women was contrary to the recommendation of increasing the consumption of low GI products. According to Androgen Excess Society recommendations, it is necessary to increase the intake of low GI products, including fruit and vegetables(Reference Conway, Dewailly and Diamanti-Kandarakis6). This approach may help to improve the effects of treatment associated with insulin resistance, which is supported by multiple intervention studies(Reference Galletly, Moran and Noakes13,Reference Phy, Pohlmeier and Cooper15–Reference Sordia-Hernandez, Rodriguez and Rodriguez17) .
In our study, the relation of DQS to BF was sought as well. We hypothesised that the diet quality, along with hormonal abnormalities, like hyperandrogemia, might support increased BF. The etiopathogenesis of obesity in PCOS has not yet been exactly explained. The hyperandrogenaemia, as was mentioned in other studies, could enhance the visceral obesity of PCOS women(Reference Allan, Strauss and Burger42,Reference Mårin, Holmtäng and Gustafsson43) . Having in mind, the fact of hormonal abnormalities in PCOS; we had a look at possible relations of BF to DQS. We could not find any relationship there. Despite this, it is necessary to underline that the comparison of PCOS and CON groups revealed that nonetheless of matching procedure for age and socio-economic status, the BF of PCOS women was higher than healthy participants. The study group had a similar body composition to subjects in other studies, further supporting the data that half of PCOS women are overweight or obese(Reference Van Hooff, Voorhorst and Kaptein44).
Considering the lower diet quality in PCOS, it is essential to find the reasons standing behind it. Keeping this in mind, we analysed the nutrition knowledge. It has been shown that knowledge influences dietary behaviours, particularly the intake of fruits and vegetables which, in turn, stimulates dietary patterns and determines BF(Reference Spronk, Kullen and Burdon45). Our study reveals that women with PCOS presented lower NK than controls. In contrast, the literature suggests that PCOS women show higher knowledge and motivation to implement preventive health strategies, which is related to the economic situation and education level(Reference Bonaccio, Di Castelnuovo and Costanzo28,Reference Beydoun and Wang46–Reference Sobas, Wadolowska and Slowinska49) . Some authors suggest that PCOS women seem to be interested in broadening their knowledge and practising healthy behaviours which are related to DQS(Reference Marijn Stok, Renner and Allan50). It is possible that women from our study were not engaged in dietary consultations but used fad diets, internet forums and were guided by advertising in their dietary choices. Additionally, it must be underlined that we have been looking at relations between NK and BF in the studied group. The current results in PCOS did not confirm the expected relationships. Such a relationship was evident in healthy controls, where the intake of prohealthy and saturated fat groups of products was driven by NK. This difference between PCOS and CON might suggest additional factors like hormonal abnormalities influencing BF in PCOS, lower motivation and dietary adherence. Additionally, such differences could be explained by nutrition knowledge or related to psychological state, which has been hypothesised by other authors and was not analysed in our study(Reference Kitzinger and Willmott51).
Although the research has reached its aims, it may have some limitations. First of all, the retrospective setting of the study. We are not sure if women already have been changing their nutrition behaviours. This problem is partially solved by the time of the diagnosis. Women in our study were diagnosed for the first time with PCOS. However, the nature of symptoms of this disorder could cause the change in their health behaviours before the study commenced. Due to this reason, and higher body weight, there is a possibility that the women involved in the study could have under-reported their intake, especially concerning unhealthy products and simple sugar sources(Reference Wardle, Haase and Steptoe52).
Second, even though the present study determined the dietary scores of 238 women by the frequency of intake of thirty-three foods, the portion size was not determined. This could have prevented an accurate assessment of the dietary nutrient intake. A lack of recording of portion sizes could explain the absence of statistical differences between the intake of non-healthy, high GI, saturated fats and dietary sugar sources. Furthermore, we did not analyse week and weekend fluctuations in dietary intake, as it was proven in previous studies, where women generally report large fluctuations in the intake of sugar and other nutrients(Reference Czlapka-Matyasik, Lonnie and Wadolowska53). Moreover, underestimation may be contributed to the compulsive eating of products generally considered unhealthy, which is not possible to observe with the FFQ(Reference Waling and Larsson54).
Third, the study did not consider the seasonality of food consumption. As the authors of previous studiesshowed, DQS would have been more precise and differentiated if the seasonality of consumption had been considered in the frequency of intake(Reference Czlapka-Matyasik and Ast55).
Fourth, the tool used to analyse the frequency of intake was validated on healthy subjects. Verification questions concerning frequency consumption were collected, analysed and revealed high repeatability of collected data. Another limitation was the size of the study sample, which restricted the use of statistical methods recommended for dietary pattern evaluation(Reference Garcia-Larsen, Morton and Norat56,Reference Cespedes and Hu57) .
Finally, considering the role of body fat distribution and cardiometabolic risk in PCOS women, it would be valuable to assess the androidal and gynoidal body fat distribution. In this study, despite the use of the ‘gold standard’, plethysmography, only total BF was assessed. Additionally, the fact that hormonal disturbances potentially influenced fat distribution should be taken into account.
Conclusions
In conclusion, our results pointed to the role of multidisciplinary therapy including dietary behaviours in PCOS women. We found differences between PCOS women and healthy controls in lower DQS related to generally healthy foods and low GI foods. PCOS women presented lower NK. The professional dietary education and improving the NK of PCOS women could affect the diet quality and increase the presence of hygienic lifestyle factors. The complementary treatment of PCOS, to ensure the optimal improvements in the disease endpoints, should be implemented after PCOS diagnosis.
Acknowledgements
Acknowledgements: The authors expressed their thanks to the participants for their contributions to the study. The authors are grateful to Dr. Katarzyna Ochmańska from Heliodor Swiecicki University Clinical Hospital for her cooperation. Financial support: The study was funded from statutory budgets of Poznan University Life Sciences (grant no.: 508-86-00). The university received a financial contribution from the Polish Ministry of Sciences and Higher Education. The authors received no financial support from commercial sources or bodies for the research, authorship and/or publication of this article. Furthermore, this research received no specific grant from any funding agency in public, commercial or not-for-profit sectors. Conflict of interest: None. Authorship: A.B.-D. contributed to the study concept and design, collected the data, performed the statistical analysis and wrote the paper. M.C.-M. was responsible for the conception and design of the study, obtained the funding, contributed to the data collection and analysis tools, wrote the paper and provided supervision and mentorship. M.K. contributed to the study concept and design, patient diagnosis and selection. K.Z. contributed to the patient diagnosis and selection and revised the manuscript for important intellectual content. M.R. revised the manuscript for important intellectual content. All authors were involved in revising the manuscript and have given their approval to the manuscript as submitted. The manuscript has been revised by all co-authors. Ethics of human subject participation:The study was conducted according to the guidelines of the Declaration of Helsinki, and all procedures involving patients were approved by the local bioethical committee of the University of Medical Sciences (acceptance number 552/16) in Poznan. Written informed consent was obtained from all patients.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980020001755
