Incorporating Objective Measures of Sedentary Behaviour Into the Detection and Control Methods of Type 2 Diabetes Mellitus in Office Employees: Development of a Mathematical Model for Clinical Practice.

doi:10.21203/rs.3.rs-3128775/v1

Download PDF

Article

Incorporating Objective Measures of Sedentary Behaviour Into the Detection and Control Methods of Type 2 Diabetes Mellitus in Office Employees: Development of a Mathematical Model for Clinical Practice.

https://doi.org/10.21203/rs.3.rs-3128775/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Introduction

: Type 2 diabetes mellitus (DM2) is one of the main public health threats of the 21st century. Identifying and predicting DM2 is the first step to stop its progression, and new strategies with low-cost, non-invasive early detection systems must be urgently implemented. Sedentary behaviour (SB) is one of the risk factors leading to the current increase in the prevalence of DM2, so incorporating the SB pattern into the detection methods of DM2 is essential.

Objective

To develop a simple mathematical model for clinical practice that allows early identification of office employees with a diagnosis of DM2 or at risk of presenting it, based on objective measurements of the SB pattern, hours of sleep and anthropometric variables.

Methods

Cross-sectional study. Anthropometric variables (sex, age and body mass index, BMI), sleep time (hours) and the SB pattern (sedentary breaks and time spent in sedentary bouts with four different lengths) of two groups of office employees (adults with and without diabetes) were measured and compared using the ActivPAL3M device. Eighty-one participants had DM2 and 132 had normal glucose metabolism (NGM). The risk of having DM2 was modelled using a generalised linear model (GLM), selecting the variables that presented a significant correlation with DM2.

Results

The DM2 prediction model used five non-invasive clinical variables -sex, age, BMI, sleep time (hours) and sedentary breaks < 20 minutes (number/day) – related to the SB pattern. The validated model correctly classified 88.89% of the participants. The model correctly classified all the office employees with NGM and 77% of office employees with DM2. It also allowed, based on the anthropometric profile of the participant, the design of a preventive tool to modify the SB pattern of office employees with DM2.

Conclusion

Understanding SB patterns by means of mathematical models could be a simple application solution for the early identification of office employees with DM2 in clinical practice. Incorporating an algorithm that contains a mathematical expression in wearable devices for monitoring the SB pattern could promote the early detection and comprehensive control of DM2 in clinical practice.

Health sciences/Endocrinology

Health sciences/Health care

Health sciences/Risk factors

Physical sciences/Mathematics and computing

Type 2 diabetes mellitus

accelerometer

diabetes

monitoring

sedentary behaviour

office employees

activPAL

lifestyle

GLM

prediction

Sedentary behaviour is a modifiable risk factor that boosts the rise of the DM2 epidemic.
This is the first study to design a mathematical model that integrates variables of the sedentary behaviour pattern to predict, control and monitor office employees with DM2 who have office jobs.
Understanding sedentary behaviour patterns by means of mathematical models could be a simple solution for the early identification and comprehensive control of DM2 in clinical practice.
Incorporating this mathematical expression in wearable devices algorithms could be a useful tool to monitor sedentary behaviour in office employees with DM2 from clinical practice.

The prevalence of diabetes has increased exponentially in recent decades, affecting 10.5% of the world population between 20 and 79 years of age in 2021, with a forecast increase to 12.2% in 2045 (1). It is estimated that almost half of all people (49.7%) living with diabetes are undiagnosed (2), with the unknown duration of the disease increasing the risks associated with it. Diabetes is responsible for 11% of annual deaths (1), and it occupies a significant economic burden in health care spending, including outpatient care, hospital care, pharmaceuticals and emergency care (3, 4). Type 2 diabetes (DM2) represents 90–95% of all cases of diabetes, and it is commonly diagnosed by glycemic measures such as fasting blood glucose concentrations of ≥ 7.0 mmol/L or glycated hemoglobin values of (HbA1c) of ≥ 6.5% (48 mmol/mol) (5). The complications caused by DM2 constitute a major worldwide public health problem, despite many pharmaceutical developments and a global emphasis on glycemic control (1), reflecting that there is a health problem in the prevention and management of diabetes both at the population and healthcare level.

The growing increase in the prevalence of DM2 coincides with the global increase in obesity (6), driven by sedentary behaviour (SB) and modern lifestyles, with increasingly sedentary office-based jobs (7). SB is one of the main reasons for the increase in the DM2 epidemic (7), in addition to other reasons such as the adoption of high energy diets in relation to low energy expenditure and population ageing (1).

SB is defined as any waking behaviour while in a sitting, reclining or lying posture characterised by an energy expenditure ≤ 1.5 times the basal metabolic rate or METs (Metabolic Equivalent Task) (8). There is contrasting evidence that people with prolonged periods of SB have an increased incidence of DM2 (9). Also, in adults with DM2, the interruption of sedentary time with brief, regular and frequent episodes of light-intensity physical activity such as walking or simple resistance activities attenuates the level of postpandrial glucose, insulin, C-peptid and triglycerides (10) and reduces waist circumference and body mass index (BMI) (11) and also blood pressure (12). DM2 could be prevented by maintaining a healthy lifestyle, engaging in physical exercise, reducing sedentary behaviour, following a healthy diet and avoiding obesity (13).

Early diagnosis and comprehensive healthcare, including management of SB, is essential to delay and stop the progression of DM2 and the evolution of associated complications. It allows the implementation of corrective measures that have been shown to be effective in reducing the morbimortality of the disease with a reduction in acute and chronic complications and of cardiovascular mortality (14).

In recent years, the approach to evaluating SB has changed from one based on the total number of hours of SB per day (total sedentary time) obtained by self-reported subjective measures (for example questionnaires, diaries) to one focused on the daily pattern of SB accumulation obtained by objective measures (for example, accelerometers). Accelerometers allow the recording and objective evaluation of the daily SB pattern in work and non-work contexts, on weekdays and at weekends, which makes it possible to characterise and differentiate the SB pattern in adults with DM2 from adults without DM2 (15) and reduce sources of potential bias inherent in the questionnaires (16). The SB pattern is defined as the manner in which SB is accumulated throughout the day while awake, including the total sedentary time, the number of sedentary interruptions/breaks daily (a non-sedentary bout between two sedentary bouts), the frequency and duration of the sedentary bouts (a period of uninterrupted sedentary time), and the time accumulated in each period (8). Compared to office employees without DM2, office employees with DM2 have a lower total number of breaks per day and present an SB pattern characterised by fewer number of breaks in time intervals of less than 20 min (sedentary breaks < 20 min) (15).

Consumer-based wearable activity trackers that allow users to objectively monitor activity levels are now widely available and may offer an alternative method for assisting individuals to remain physically active (17) and reduce SB. Currently, most of these devices do not measure the SB pattern characteristic in adults with DM2 (sedentary breaks < 20 min) and therefore, cannot be used in clinical practice as objective tools to identify adults with DM2. Developing predictive mathematical models that contain variables of the SB pattern would provide a simple, easily obtainable, objective and inexpensive tool to detect a greater number of adults with DM2 early, maximising the effectiveness for the prevention and control of the disease.

The main objective of this study was to develop a simple mathematical model –with variables that are easy to measure outside the consultation and are non-invasive– that allows the identification of office employees with DM2 or at risk of suffering from it, using measurements of the SB pattern. Having a good model would enable primary healthcare professionals to (i) take preventive measures and achieve early detection of DM2 in consultations, and (ii) develop an easy-to-use, personalised tool, adapted to anthropometric characteristics, to modify the SB pattern of office employees with DM2 from clinical practice.

2.1. Study design

A cross-sectional study comparing sex, age, anthropometric variables (BMI) and the SB pattern of two groups of participants, office employees with DM2 and office employees with normal glucose metabolism (NGM).

2.2. Participants

The inclusion criteria for the group of office employees with DM2 was to be diagnosed with DM2 in accordance with international criteria (18). The rest of the inclusion criteria were common for both groups: adults between 18 and 65 years old (working age); office employees with a minimum of 55% of their daily working hours performing sedentary tasks according to the Occupational Sitting and Physical Activity Questionnaire (OSPAQ) (19) and; having a work contract of at least 18.5 hours/week. The exclusion criteria for both groups consisted of having a diagnosis of musculoskeletal, cardiovascular, pulmonary or orthopaedic problems or any other physical condition that prevented them from being physically active.

2.3. Setting

The recruitment was done in the framework of two clinical trials. The group of office employees with NGM were a convenience sample of administrative staff (n = 132) recruited from four Spanish hospitals (April to December 2016) (20). The study was approved by the ethics committee of each hospital (20), and the recruitment procedure is detailed elsewhere (20). Briefly, the Occupational Health Services from each hospital contacted all their administrative staff (n = 321) and invited them to participate in the study. Two hundred and fifteen employees expressed interest in participating. Employees were considered eligible if they had access to a mobile phone with an Android version > 4.0, had no physical or health problems that limited their ability to stand for bouts of at least 10 min, and had no planned absence from work for more than one week over the next two months (20). Of these, 73 were excluded due to failing to meet the inclusion criteria. A total sample of 142 administrative staff were finally recruited. Ten participants were additionally excluded for errors in the record on activPAL3TM based-data.

The group of office employees with DM2 (n = 81) was recruited at primary care centres in Barcelona between April 2019 and January 2020. In each centre and during the recruitment process, family physicians and nurses identified and recruited patients in medical consultations through the health centre’s database after verifying that they met the inclusion criteria. We contacted 160 office employees with DM2 that met the inclusion criteria. Forty-two office employees with DM2 did not respond, 20 were not interested and 17 failed to record correctly the measures taken by the activPAL3TM device (i.e. insufficient record of days or errors in the record). The total number of office employees with DM2 with the record of minimum days set was 81. Written informed consent was obtained from all subjects included in the study. This trial was approved by the IDIAP Jordi Gol Clinical Research Ethics Committee with registration number P18/102. The recruitment procedure is detailed elsewhere (21).

2.4. Variables and data measurement

The activPAL3TM device (PAL Technologies Ltd., Glasgow, UK) measured and quantified for 24 hours per day for seven complete days (Monday to Sunday) the physical activity, SB patterns and sleep time of office employees with DM2 and NGM. This device is a valid measure to quantify body posture and activity patterns during free-living conditions (22).

From the activPAL3TM software output, the following SB variables were determined as they represented the SB pattern characteristic of office employees with DM2 (15): sleep time (hours), total number of sitting bouts and number of sitting bouts with four different lengths (< 20 min, 20–40 min, 40–60 min and > 60 min). Overall, the outcomes related to the SB pattern were reported as averages of weekdays (working and non-working hours) and weekend days, taking into account that SB patterns can vary depending on whether they happen during weekends or weekdays and whether or not there are work commitments (working hours vs non-working hours) (23). Working and non-working times were established by using the participants’ daily paper records.

Age, sex and anthropometric variables by weight, height and BMI were also recorded for each adult. The anthropometric variables were measured with a Seca 770 scale and a Seca 222 height measuring rod. The BMI was calculated by dividing the body weight by the square of the height in metres (kg/m2).

2.5. Sample size

The sample was 213 patients. The sample size was initially set at 132 patients with DM2 and 132 adults with NGM. The total number of adults with NGM and DM2 with the record of minimum days set was 132 and 81 respectively. Using the free software G*Power (24) for a comparison test of bilateral means and a significance level of 5%, a power of 95% was obtained.

2.6. Statistical methods

A model is useful if it is simple, capable of explaining the behaviour of the variable of interest (principle of parsimony), technically congruent and interpretable (25). An increase in the number of variables implies higher standard errors. All those variables that are considered technically important should be included in a model. First, we selected the variables that could potentially better explain the presence or absence of DM2 according to the useful model requirements (15). Second, and to build the prediction model, we introduced these variables (15) in the GLM model. We then studied if there was a correlation between the selected variables to avoid subsequent collinearity problems in the models. Pearson’s correlation coefficients were calculated between the variables that measured the SB pattern, anthropometric variables and sleep hours with the variable presenting or nor presenting DM2. If highly correlated, explanatory variables were introduced in the construction of GLM and a multicollinearity problem was generated, which seriously affected the estimators obtained and their precision. Then, we built the prediction model and, we introduced these variables (15) in the GLM model.

The SB pattern was measured in four units of time in the week: weekday total time, weekday during working time, weekday during non-working time and weekend. The GLM models were studied for each of these time units, taking the presence of DM2 as the response variable and those selected as using the Pearson correlation coefficient as the independent variables.

The response variable of the model was in every case dichotomous, being able to take two values: presenting DM2 (0) or not presenting DM2 or NGM (1). This binary variable was explained using continuous variables related to the SB pattern, age, sex, and anthropometric variables (weight and BMI). The GLM that was used had a binomial error distribution and a logit link function. The following logistic regression was performed:

$$logit\left(DM2\right)={\alpha }_{0}+\sum _{}^{}{\alpha }_{i}\bullet {X}_{i}$$

$$P\left(DM2\right)=\frac{1}{1+{e}^{-logit\left(Not DM2\right)}}$$

The response of the model resulted in a value between 0 and 1, interpreted as the probability that the office employee did not have DM2. If the value was less than 0.5, it was accepted that the office employee had DM2; and if it was greater than 0.5, it was accepted that the office employee did not have DM2. The values in the central band of the 0–1 interval were interpreted with some caution since in these cases the model predicted, approximately, a probability of 0.5 of presenting DM2 and of 0.5 of not presenting DM2. The central band was defined as the interval of 0.4–0.6. Participants in these intervals are recommended to take the conventional blood tests (HbA1c, glucose) in order to confirm identification as office employees with DM2. To do the statistical treatment, the free software R, R Core Team (2022) was used (26).

Data were gathered from 80 men and 131 women; 72.5% of the men presented DM2 vs 16% of the women. The mean age of the office employees with DM2 was 10.6 years older than office employees with NGM (p < 0.001). The BMI of the office employees with DM2 was 6.5 units higher than office employees with NGM (p < 0.001) (15).

3.1. Pearson correlation coefficient

In the four units of time in the week (weekday total time, weekday working time, weekday non-working time and weekend), the variables age, BMI, sex, sleeping time, sedentary breaks < 20min (number/day), time spent in sedentary bouts < 20 min (minutes/day) and total sedentary breaks (number/day) were variables that significantly correlated with the variable presenting DM2 (see Table 1). In the case of age, BMI and sleeping time, the correlation was positive, increasing the values of these variables in office employees with DM2. The correlation was negative for the three variables that measured the SB pattern, and therefore the office employees with DM2 had lower values. In the case of gender, the correlation was negative, increasing diabetes in men (see Table 1).

Table 1

Pearson’s correlation coefficient and p-value between the recorded variables and the variable presenting DM2 in the four units of time sampled.
	Weekday -Total time		Weekend
	correlation	p-value	correlation	p-value
Age (years)	0.5439	< .0001	0.537	< .0001
BMI (kg/m²)	0.5255	< .0001	0.539	< .0001
Sex	-0.5738	< .0001	-0.5538	< .0001
Sleep time (hours)	0.3956	< .0001	0.1981	0.0049
Light-intensity physical activity (minutes/day)	-0.0036	0.9585	-0.1342	0.0581
Moderate to vigorous physical activity (minutes/day)	-0.0915	0.1907	0.0311	0.6624
Sedentary breaks < 20 min (number/day)	-0.5059	< .0001	-0.3873	< .0001
Sedentary breaks 20–40 min (number/day)	0.0479	0.4937	0.0619	0.384
Sedentary breaks 40–60 min (number/day)	0.1614	0.0205	0.1195	0.092
Sedentary breaks > 60 min (number/day)	0.1187	0.0892	0.0988	0.1641
Time spent in sedentary bouts < 20 min (minutes/day)	-0.4431	< .0001	-0.1092	0.1239
Time spent in sedentary bouts 20–40 min (minutes/day)	0.051	0.467	0.087	0.2207
Time spent in sedentary bouts 40–60 min (minutes/day)	0.1471	0.0348	0.1123	0.1132
Time spent in sedentary bouts > 60 min (minutes/day)	0.126	0.071	0.0847	0.2329
Total sedentary breaks (number/day)	-0.5239	< .0001	-0.4005	< .0001
Total time spent in sedentary bouts (minutes/day)	-0.1064	0.1278	0.1045	0.1408
	Weekday – Working time		Weekday Non-Working time
	correlation	p-value	correlation	p-value
Age (years)	0.4831	< .0001	0.4831	< .0001
BMI (kg/m²)	0.4209	< .0001	0.4209	< .0001
Sex	-0.5541	< .0001	-0.5541	< .0001
Sleep time (hours)
Light-intensity physical activity (minutes/day)	0.3413	< .0001	-0.258	0.0011
Moderate to vigorous physical activity (minutes/day)	0.1379	0.0839	-0.0231	0.7733
Sedentary breaks < 20 min (number/day)	-0.3016	0.0001	-0.443	< .0001
Sedentary breaks 20–40 min (number/day)	-0.0498	0.5345	0.0141	0.8601
Sedentary breaks 40–60 min (number/day)	0.1278	0.1095	-0.0034	0.966
Sedentary breaks > 60 min (number/day)	0.1694	0.0333	-0.0927	0.2468
Time spent in sedentary bouts < 20 min (minutes/day)	-0.2756	0.0005	-0.3563	< .0001
Time spent in sedentary bouts 20–40 min (minutes/day)	-0.0353	0.6596	-0.0043	0.9568
Time spent in sedentary bouts 40–60 min (minutes/day)	0.1173	0.1422	-0.0158	0.8439
Time spent in sedentary bouts > 60 min (minutes/day)	0.2074	0.0089	-0.0716	0.3711
Total sedentary breaks (number/day)	-0.3156	< .0001	-0.452	< .0001
Total time spent in sedentary bouts (minutes/day)	-0.0226	0.7783	-0.2265	0.0042
BMI: body mass index

The variables of the SB pattern correlated with DM2 were also significantly correlated with each other (see Table 2). To avoid collinearity, only the variable that measured sedentary breaks < 20min (number/day) was used in the GLM model, discarding the variables time spent in sedentary bouts < 20 min (minutes/day) and total sedentary breaks (number/day). The variable that measured the average hours/day of sleep on weekdays was not correlated with the variables sedentary breaks < 20min (number/day), time spent in sedentary bouts < 20 min (minutes/day) and total sedentary breaks (number/day).

Table 2

Pearson’s correlation coefficient and p-value between the variables related to SB that are correlated with the DM2 variable.
Pearson’s Correlation (p-value)	Sedentary breaks < 20 min (number/day)	Time spent in sedentary bouts < 20 min (minutes/day)	Total sedentary breaks (number/day)
Weekday Total time
Sleep time (hours)	-0.0995 (0.1508)	-0.2016 (0.0033)	-0.1339 (0.0527)
Sedentary breaks < 20 min (number/day)		0.7711 (< 0.0001)	0.9948 (< 0.0001)
Time spent in sedentary bouts < 20 min (minutes/day)			0.7815 (< 0.0001)
Weekend
Sleep time (hours)	-0.3182 (< 0.0001)	-0.3165 (< 0.0001)	-0.3424(< 0.0001)
Sedentary breaks < 20 min (number/day)		0.7331 (< 0.0001)	0.9948 (< 0.0001
Time spent in sedentary bouts < 20 min (minutes/day)			0.737 (< 0.0001)
Weekday – Non-Working time
Sleep time (hours)	0.0482 (0.5439)	0.1183 (0.1351)	0.082 (0.3009)
Sedentary breaks < 20 min (number/day) weekday-		0.661 (< 0.0001)	0.9954 (< 0.0001)
Time spent in sedentary bouts < 20 min (minutes/day)			0.669 (< 0.0001)
Weekday - Working time
Sleep time (hours) weekend	-0.0221 (0.7806)	0.0786 (0.3217)	-0.0128 (0.8717)
Sedentary breaks < 20 min (number/day) weekday-working time		0.8357 (< 0.0001)	0.9953 (< 0.0001)
Time spent in sedentary bouts < 20 min (minutes/day) weekday-working time			0.8495 (< 0.0001)

3.2. General linear model (GLM)

Taking into account the variables significantly correlated with the presence of DM2, the parameters of the following model were estimated:

$$logit\left(DM2\right)={\alpha }_{0}+{\alpha }_{1}\bullet Age+{\alpha }_{2}\bullet BMI+{\alpha }_{3}\bullet Sex +{\alpha }_{4}\bullet Sleep+{\alpha }_{5}\bullet Breaks20$$

In the models corresponding to weekdays, the variable sleep time was introduced since it was significantly correlated in this period of time. The model that used the information on weekday working time and weekday non-working time showed the greatest deviance explaining at least 10% more of the variability of the data compared to the other models (see Table 3). If the sleep time variable was eliminated, the deviance was 40.27%. If the model included the information on the SB pattern of weekdays and weekends simultaneously, the variables corresponding to holidays were not significant.

Table 3

Parameters and deviance of the GLM models for different units of time.
Model		Estimate	Std. Error	Pr(>\|z\|)
Weekday – Total time Deviance 66.48%	(Intercept)	-17.18	4.41	< 0.001	***
	Age (years)	0.17	0.04	< 0.001	***
	BMI (Kg/m²)	0.16	0.06	0.004	**
	Sex	-2.22	0.55	< 0.001	***
	SleepWeek (hours)	1.37	0.37	< 0.001	***
	Sedentary breaks < 20min (number/day)	-0.06	0.02	< 0.001	***
Weekend Deviance 53.46%	(Intercept)	-8.49	2.58	< 0.001	***
	Age (years)	0.14	0.03	0.000	***
	BMI (Kg/m²)	0.18	0.05	0.000	***
	Sex	-2.24	0.49	0.000	***
	Sedentary breaks < 20min (number/day)	-0.03	0.02	0.077
Weekday - Working time Deviance 56.76%	(Intercept)	-18.41	4.66	< 0.001	***
	Age (years)	0.17	0.05	< 0.001	***
	BMI (Kg/m²)	0.15	0.06	0.013	*
	Sex	-2.39	0.58	< 0.001	***
	SleepWeek (hours)	1.34	0.39	0.001	***
	Sedentary breaks < 20min (number/day)	-0.04	0.02	0.024	*
Weekday – Non-Working time Deviance 60.17%	(Intercept)	16.45	4.21	0.013	*
	Age (years)	0.14	0.05	< 0.003	**
	BMI (Kg/m²)	0.12	0.05	0.040	*
	SleepWeek (hours)	-1.25	0.33	< 0.001	***
	Sedentary breaks < 20min (number/day)	-0.11	0.04	< 0.004	**
	Sex	-2.46	0.39	< 0.001	***
BMI: body mass index

The model chosen to make predictions was the one that used the significant variables in weekday total time. The variables that intervened in the selected model (1) were sex, age, the average number of hours of sleep (sleep time), the BMI and the number of breaks in time intervals of less than 20 minutes.

$$logit\left(DM2\right)=-17.18+0.17\bullet Age+0.16\bullet BMI-2.22\bullet Sex+1.37\bullet Sleep-0.06 \bullet WB20.$$

In a logistic regression, the interpretation of the model’s coefficients is not directly quantifiable. However, positive coefficient values indicated in our model an increase in the probability of presenting DM2. We can therefore say that an increase in age, BMI and sleep time increased the probability of presenting DM2. While gender (being a female) and an increase in the number of breaks of less than 20 minutes reduced de risk of presenting DM2.

3.3. Validation of the mathematical model

To carry out the validation, 30 random numbers between 1 and 213 were generated. To estimate the parameters of the model, the patients who took up the position of the 30 numbers were removed from the sample. The model obtained has been validated with the 30 participants who had not taken part in the construction of the mathematical model. The model correctly classified 88.89% of the patients. It classified all the office employees with NGM correctly and 77% of office employees with DM2. Three women were erroneously identified as office employees with NGM: one was 40 years of age, another had a very low BMI (19.71 kg/m²) and slept an average of six hours a day, and the third was a 51-year-old woman with a BMI of 38.77 kg/m² who performed an average number of 63.5 sedentary breaks of less than 20 minutes. The model predicted a mean probability of 0.89 of suffering from DM2 for office employees diagnosed with the disease and a mean probability of 0.18 for office employees with NGM.

3.4. Recommendation of the number of sedentary breaks less than 20min (number/day)

To develop the practical tool, we fixed the values of the non-modifiable variables in the short term and required the model to take a value greater than 0.5 (probability of presenting NGM), a value that would classify the individual as non-diabetic from which it was possible to obtain an estimate of the minimum number of breaks < 20 minutes required to modify the SB pattern of office employees with DM2 (see Table 4).

Table 4

Number of breaks < 20 minutes during sedentary behaviour recommended according to the proposed mathematical model (1)
BMI (kg/m²)	Age (years)	Sleep time (hours)
		Men			Women
		7 hours	7.5 hours	8 hours	7 hours	7.5 hours	8 hours
Normal	50	28	39	50	9	2	13
Normal	60	56	67	79	19	30	42
Overweight 25-29.9 kg/m²	40	17	28	39	-	5	2
	50	45	56	68	8	19	31
	60	73	85	96	36	48	59
Obese 30-39.9 kg/m²	40	30	41	53	20	4	16
	50	58	70	81	21	33	44
	60	87	98	109	50	61	72

The BMI variable was classified into three groups: normal, overweight and obese, in accordance with current guidelines from the World Health Organization (WHO), which defines a normal BMI as 18.5 to 24.9 kg/m². A BMI ≥ 25 kg/m² is considered to be overweight, a BMI ≥ 30 kg/m² obese, and severe obesity is defined as a BMI ≥ 40 kg/m² (27).

The results of Table 4 could allow health professionals to estimate the minimum number of breaks during sedentary time required to modify the SB pattern, adapted to office employees with DM2 according to sex, age, and BMI and sleep time. For example, for a 60-year-old office employee with a BMI of 30 kg/m² who sleeps eight hours a day, the minimum number of breaks < 20 minutes per day would be 109, while if the office employee were a woman, it would be 72 breaks/day.

Through the development of a mathematical model, this study contributed to identifying, with a high level of accuracy (88.89%), office employees with DM2, based on a measurement of the SB pattern (number of breaks < 20 minutes/day), sleep time (hours), sex, age and BMI. These factors are easy to measure with non-invasive methods, are known to be associated with the risk of DM2 (9–11) and focus attention on new approaches and methods for the early detection of DM2 from clinical practice.

Through the mathematical model a clinical tool that was individualised for office employees with DM2, easy to use, practical and with no increase in health cost was designed, in order to assess the number of daily sedentary breaks. It can provide primary healthcare professionals with predictive information so as to integrate the interventions of prevention and modifications of the SB pattern and can be used routinely in clinical practice to encourage the performance of confirmatory tests for the diagnosis of DM2 in a greater number of patients. Since diabetes is an underdiagnosed disease, with almost half of those with diabetes being unaware they have it or asymptomatic (28), the use of this predictive tool could have a great impact at the health, social and economic level by being able to predict the evolution of this disease and minimise its consequences in the population.

Most of the existing DM2 prediction models are based on a large number of laboratory variables (glucose levels, HbA1c, uric acid, blood osmolality, sodium, blood urea nitrogen, triglycerides, LDL cholesterol, among others) that are costly to obtain, involve invasive methods that require blood tests in the laboratory (29) and are not available in the general population, not even in developed countries. In addition, models for classifying diabetes that include non-laboratory variables (for example, sex, drinking, smoking, physical activity) do not take into account objective characteristics of the SB or physical activity pattern, since they are based on self-reports or subjective methods (29, 30). The mathematical model developed in this study requires basic information within the reach of the majority of the population, with fewer geographical limitations and with a potential improvement of cost-efficiency compared to traditional models.

The mathematical model, based on objective measurements of the SB pattern, supports the need to take these variables into account since there is robust evidence that SB is one of the main risk factors that explain the rise in the DM2 epidemic (7, 9, 10, and 31). Taking advantage of the rise of mobile technology and the growth in the market of devices for the objective measurement of physical activity and SB, introducing measures of the daily number of breaks in sitting time in these devices could provide the necessary data to use the model in healthcare practice (17). Wearable activity trackers are a low-cost tool, have a wide scope, are clinically beneficial and can improve physical activity and SB levels in the clinical population (32) in the long term. In addition, they provide automated feedback on the time spent in different intensities of physical activity and sedentary behaviours, making them interactive tools for behaviour change through smartphone apps (33). Incorporating an algorithm that contains the model in wearable devices to change the SB pattern can enable a person at risk to take measures that can prevent the development or delay the progression of the disease by increasing the number of breaks during the day.

The main strength of this study includes the objective measures of sedentary behaviour in all contexts, including in the workplace and during leisure time. The instrument (the ActivPal device) used to measure the SB pattern has shown good reliability and validity and is considered the gold standard for measuring SB, especially for measuring the accumulation of sedentary time throughout the day (22). The second strength is the representativeness of the study sample regarding office employees with DM2. The prevalence of DM2 is higher in men than women and increases with age (28). The sample of office employees living with diabetes also had a lower proportion of women, tended to be older and had a higher BMI, in line with other studies (34, 35). The study sample also slept for long hours, similar to what is generally observed in adults with DM2 who sleep for many hours but of worse quality, contributing to the development of comorbidities such as obesity or sleep obstructive disorder (36). The study sample focused in office employees with DM2 because most adults in Western nations spend most of their days employed (37) and have desk-based jobs in which sitting is the default (38). Office employees with DM2 were treated with both oral anti-diabetic medication and insulin and were monitored and controlled by primary health care. Given that patients with DM2 visit primary care providers frequently (39), it is especially important to focus the interventions on office employees (jobs with little physical demand) given the magnitude and differentiated risk of a sedentary lifestyle in this type of administrative jobs since they are the ones that report the highest prevalence of SB (40). The main limitation of this study is that the analysis was cross-sectional, therefore causal relationships could not be examined. In addition, future research should consider the generalization of this mathematical model to all adults with DM2.

This is the first study to design a mathematical model that integrates variables of the SB pattern to predict, control and monitor office employees with DM2 from clinical practice. Future research should consider the use of objective, non-invasive variables such as the number of sedentary breaks under 20 minutes/day in the prediction models and management tools for DM2 in clinical practice. It is necessary to include this variable since it is characteristic of office employees with DM2 (15) and it is a modifiable factor, so it is essential to design SB intervention tools at the level of clinical practice for the effective control of the disease.

DM2: Type 2 diabetes mellitus

SB: sedentary behaviour

NGM: normal glucose metabolism

BMI: body mass index

GLM: generalised linear model

HbA1c: Haemoglobin A1c

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and approved by the Clinical Research Ethics Committee of the Primary Care Research Institute Jordi Gol i Gurina with the registration code P18/102 and date of approval 24 October 2018. Informed consent was obtained from all subjects involved in the study.

Consent for publication

Not applicable.

Availability of data and materials

Pertinent data generated or analysed as a result of this study protocol in the future will be made available in publications, and some portions could be made available from the corresponding author upon reasonable request.

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

The study was funded by the Fondo de Investigación Sanitaria (Fund for Research in Health Sciences), Instituto de Salud Carlos III (PI17/01788), the Spanish Ministry of Science and Innovation (DEP2021-37169), a Jose Luís Torres grant from redGDPS Foundation (redgdps/BJLTB/02/2022), the predoctoral research grant Isabel Fernández 2020 from the Spanish Society of Family and Community Medicine (semFYC), and another grant from the Càtedra Autonomous University of Barcelona (UAB)– Novartis for research in Family and Community Medicine. The funders had no role in the design, analysis, data interpretation or writing of the manuscript.

Authors' contributions

FA, MAC, JBR, CPE, CZY, CMC and APR were responsible for the conception and design of the study. MAC was responsible for the sample size calculation and conceived the statistical methods. FA, APR, JBR, CZY and CPR participated in the development of the study. All authors have performed a critical revision of this manuscript and the final version.

Acknowledgements

We thank the Foundation University Institute for Primary Health Care Research Jordi Gol i Gurina for supporting this work.

Authors' information

^aPrimary Healthcare Centre Passeig de Sant Joan, Catalan Health Institute, Barcelona, Spain; ^bMember of the redGDPS Foundation, Madrid, Spain; ^cSport and Physical Activity Research Group, Centre for Health and Social Care Research, University of Vic-Central University of Catalonia, Vic, Spain; ^dResearch Group on Methodology, Methods, Models and Outcomes of Health and Social Sciences, Centre for Health and Social Care Research, University of Vic-Central University of Catalonia, Barcelona, Spain; ^eBarcelona Research Support Unit, Primary Care Research Institute IDIAP Jordi Gol, Barcelona, Spain; ^fCIBER of Diabetes and Associated Metabolic Disease (CIBERDEM). Instituto de Salud Carlos III (ISCIII). Madrid, Spain; ^gDepartment of Mathematics, ETSEA, University of Lleida, Lleida, Spain.

Conflict of Interest

None

Sun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119. doi:10.1016/j.diabres.2021.109119
Cho NH, Shaw JE, Karuranga S, et al. IDF Diabetes Atlas: Global estimates of diabetes prevalence for 2017 and projections for 2045. Diabetes Res Clin Pract. 2018;138:271-281. doi:10.1016/j.diabres.2018.02.023
Dieleman JL, Baral R, Birger M, Bui AL, Bulchis A, et al. 2016. US spending on personal health care and public health, 1996–2013. JAMA 316:2627–46
Bommer, Heesemann, E, Sagalova, V, et alThe global economic burden of diabetes in adults aged 20-79 years: a cost-of-illness study. Lancet Diabetes Endocrinol 2017; 5(6): 423–430.
American Diabetes Association 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2020 . Diabetes Care 2020;43(Suppl 1):S14-31. 10.2337/dc20-S002
Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nat Rev Endocrinol 2018;14:88-98. 10.1038/nrendo.2017.151
Rockette-Wagner B, Edelstein S, Venditti EM, Reddy
D, Bray GA, Carrion-Petersen ML, et al. The impact of lifestyle intervention on sedentary time in individuals at high risk of diabetes. Diabetologia 2015; 58 (6): 1198-
Tremblay MS, Aubert S, Barnes JD, et al. Sedentary Behavior Research Network (SBRN) - Terminology Consensus Project process and outcome. Int J Behav Nutr Phys Act. 2017;14(1):75. Published 2017 Jun 10. doi:10.1186/s12966-017-0525-8
Wilmot EG, Edwardson CL, Achana FA, et al. Sedentary time in adults and the association with diabetes, cardiovascular disease and death: systematic review and meta-analysis [published correction appears in Diabetologia. 2013 Apr;56(4):942-3]. Diabetologia. 2012;55(11):2895-2905. doi:10.1007/s00125-012-2677-z
Loh R, Stamatakis E, Folkerts D, Allgrove JE, Moir HJ. Effects of Interrupting Prolonged Sitting with Physical Activity Breaks on Blood Glucose, Insulin and Triacylglycerol Measures: A Systematic Review and Meta-analysis. Sports Med. 2020;50(2):295-330. doi:10.1007/s40279-019-01183-w
Healy GN, Winkler EA, Brakenridge CL, Reeves MM, Eakin EG. Accelerometer-derived sedentary and physical activity time in overweight=obese adults with type 2 diabetes: cross-sectional associations with cardiometabolic biomarkers. PLoS One. 2015;10(3):e0119140.
Dempsey PC, Larsen RN, Dunstan DW, Owen N, Kingwell BA. Sitting Less and Moving More: Implications for Hypertension. Hypertension. 2018;72(5):1037-1046. doi:10.1161/HYPERTENSIONAHA.118.11190
Hu FB, Manson JE, Stampfer MJ, Colditz G, Liu S, Solomon CG, et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med 2001 Sep 13;345(11):790-797.
American Diabetes Association. 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes-2021 [published correction appears in Diabetes Care. 2021 Sep;44(9):2182]. Diabetes Care. 2021;44(Suppl 1):S15-S33. doi:10.2337/dc21-S002
Colomer, F.A, Cugat, M.À.C, Bort-Roig, J, et al. Differences in Free-Living Patterns of Sedentary Behaviour between Office Employees with Diabetes and Office Employees without Diabetes: A Principal Component Analysis for Clinical Practice. Int. J. Environ. Res. Public Health 2022, 19, 12245. https://doi.org/10.3390/ijerph191912245
Lee IM, Shiroma EJ. Using accelerometers to measure physical activity in large-scale epidemiological studies: issues and challenges. Br J Sports Med. 2014;48(3):197–201.
Brickwood KJ, Watson G, O'Brien J, Williams AD. Consumer-Based Wearable Activity Trackers Increase Physical Activity Participation: Systematic Review and Meta-Analysis. JMIR Mhealth Uhealth. 2019;7(4):e11819. Published 2019 Apr 12. doi:10.2196/11819
American Diabetes Association. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes. Diabetes Care 2020;43(Suppl 1):S14-S31. doi:10.2337/dc20-S002
Chau JY, Van Der Ploeg HP, Dunn S, Kurko J, Bauman AE. Validity of the occupational sitting and physical activity questionnaire. Med Sci Sports Exerc 2012;44(1):118-125. doi:10.1249/MSS.0b013e3182251060
Bort-Roig J, Chirveches-Pérez E, Giné-Garriga M, Navarro-Blasco L, Bausà-Peris R, Iturrioz-Rosell P, González-Suárez AM, Martínez-Lemos I, Puigoriol-Juvanteny E, Dowd K, Puig-Ribera A. An mHealth Workplace-Based "Sit Less, Move More" Program: Impact on Employees' Sedentary and Physical Activity Patterns at Work and Away from Work. Int J Environ Res Public Health. 2020 Nov 28;17(23):8844. doi: 10.3390/ijerph17238844. PMID: 33260697; PMCID: PMC7730175.
Alòs F, Colomer MÀ, Martin-Cantera C, et al. Effectiveness of a healthcare-based mobile intervention on sedentary patterns, physical activity, mental well-being and clinical and productivity outcomes in office employees with type 2 diabetes: study protocol for a randomized controlled trial. BMC Public Health. 2022;22(1):1269. Published 2022 Jun 29. doi:10.1186/s12889-022-13676-x
Dowd KP, Harrington DM, Bourke AK, Nelson J, Donnelly AE. The measurement of sedentary patterns and behaviors using the activPAL™ Professional physical activity monitor. Physiol Meas 2012;33(11):1887-1899. doi:10.1088/0967-3334/33/11/1887
Burton NW, Haynes M, van Uffelen JG, Brown WJ, Turrell G. Mid-aged adults' sitting time in three contexts. Am J Prev Med. 2012;42(4):363-373. doi:10.1016/j.amepre.2011.11.012
Faul F, Erdfelder E, Lang AG, Buchner A. G. Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007;39(2):175-191. doi:10.3758/bf03193146.
Vandekerckhove, JS, Matzke, D., Wagenmakers, E: Model Comparison and the Principle of Parsimony; Oxford Handbook of Computational and Mathematical Psychology. OxfordLibrary of Psychology, 2015. Available online: https://escholarship.org/uc/item/9j47k5q9 (accesed on 1 October 2022).
R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2022; Available online: https://www.R-project.org/ (accessed on 1 October 2022).
World Health Organization. Obesity and overweight. Fact sheet N°311 January 2015. Available online: http://www.who.int/mediacentre/factsheets/fs311/en/ (accessed on 15 September 2022).
Soriguer F, Goday A, Bosch-Comas A, et al. Prevalence of diabetes mellitus and impaired glucose regulation in Spain: the [email protected] Study. Diabetologia. 2012;55(1):88-93. doi:10.1007/s00125-011-2336-9
Dinh A, Miertschin S, Young A, Mohanty SD. A data-driven approach to predicting diabetes and cardiovascular disease with machine learning. BMC Med Inform Decis Mak. 2019;19(1):211. Published 2019 Nov 6. doi:10.1186/s12911-019-0918-5
Deberneh HM, Kim I. Prediction of Type 2 Diabetes Based on Machine Learning Algorithm. Int J Environ Res Public Health. 2021;18(6):3317. Published 2021 Mar 23. doi:10.3390/ijerph18063317
González K, Fuentes J, Márquez JL. Physical Inactivity, Sedentary Behavior and Chronic Diseases. Korean J Fam Med. 2017;38(3):111-115. doi:10.4082/kjfm.2017.38.3.111
Ferguson T, Olds T, Curtis R, et al. Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses. Lancet Digit Health. 2022;4(8):e615-e626. doi:10.1016/S2589-7500(22)00111-X
Schoeppe S, Alley S, Van Lippevelde W, et al. Efficacy of interventions that use apps to improve diet, physical activity and sedentary behaviour: A systematic review. Int J Behav Nutr Phys Act 2016; 13: 127
Vézina-Im LA, Morin CM, Desroches S. Sleep, Diet and Physical Activity Among Adults Living With Type 1 and Type 2 Diabetes. Can J Diabetes. 2021;45(7):659-665. doi:10.1016/j.jcjd.2021.01.013
Menéndez Torre EL, Ares Blanco J, Conde Barreiro S, Rojo Martínez G, Delgado Alvarez E; en representación del Grupo de Epidemiología de la Sociedad Española de Diabetes (SED). Prevalence of diabetes mellitus in Spain in 2016 according to the Primary Care Clinical Database (BDCAP). Prevalencia de diabetes mellitus en 2016 en España según la Base de Datos Clínicos de Atención Primaria (BDCAP). Endocrinol Diabetes Nutr (Engl Ed). 2021;68(2):109-115. doi:10.1016/j.endinu.2019.12.004
Jehan S, Myers AK, Zizi F, Pandi-Perumal SR, Jean-Louis G, McFarlane SI. Obesity, obstructive sleep apnea and type 2 diabetes mellitus: Epidemiology and pathophysiologic insights. Sleep Med Disord. 2018;2(3):52-58. Epub 2018 Jun 21. PMID: 30167574; PMCID: PMC6112821.
Ausubel, J.H. & Grubler, A. (1995). Working less and living longer: Long-term trends in working time and time budgets. Technological Forecasting and Social Change 50 (3) 195-213.
Buckley JP, Hedge A, Yates T, et al. The sedentary office: an expert statement on the growing case for change towards better health and productivity. Br J Sports Med. 2015;49(21):1357-1362. doi:10.1136/bjsports-2015-094618
Grant RW, McCarthy EP, Singer DE, Meigs JB. Frequent outpatient contact and decreasing medication affordability in patients with diabetes from 1997 to 2004. Diabetes Care. 2006;29(6):1386-1388. doi:10.2337/dc06-0196
Vandelanotte C, Duncan MJ, Short C, Rockloff M, Ronan K, Happell B, Di Milia L. Associations between occupational indicators and total, work-based and leisure-time sitting: a cross-sectional study. BMC Public Health. 2013 Dec 1;13:1110. doi: 10.1186/1471-2458-13-1110. PMID: 24289321; PMCID: PMC3879072.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Incorporating Objective Measures of Sedentary Behaviour Into the Detection and Control Methods of Type 2 Diabetes Mellitus in Office Employees: Development of a Mathematical Model for Clinical Practice.

Status:

Version 1

Abstract

Introduction

Objective

Methods

Results

Conclusion

Contributions to the Literature

1. Background

2. Methods

2.1. Study design

2.2. Participants

2.3. Setting

2.4. Variables and data measurement

2.5. Sample size

2.6. Statistical methods

3. Results

3.1. Pearson correlation coefficient

3.2. General linear model (GLM)

3.3. Validation of the mathematical model

3.4. Recommendation of the number of sedentary breaks less than 20min (number/day)

4. Discussion

4. Conclusions

Abbreviations

Declarations

References

Additional Declarations

Status:

Version 1