Correlates of problematic internet use among undergraduate medical students of Delhi

The use of the internet has increased dramatically over the past two decades. Recent global internet statistics by global reach show over 803 million people have online access worldwide [1]. The benefits of the internet have been widely researched and include education and research, communication, health-related services, online monetary transactions, trade, buying goods, entertainment, etc. [2]. Researchers found that 73% of college students accessed the internet at least once a day and spent approximately 1.6 to 4.5 h a day online, preferably during the night [2‐5]. Globally, due to population diversity, different instruments, cut-off scores used, and different sample characteristics, the prevalence of problematic internet use (PIU) vary from 7.3 to 51% [6‐11]; whereas, In India, it ranges from 7.45 to 19.85% among undergraduate medical students [2, 3, 12, 13].

Globally, deteriorating effects of problematic internet use were studied by various researchers; they found that PIU was associated with a wide range of social and psychological problems, like academic failure, poor self-confidence, poor psychological well-being, sleep deprivation, social withdrawal, poor diet, and cardiopulmonary complications [13‐15]. The mean scores of the following domains: anxiety, depression, paranoid ideation, and obsessive-compulsive of Symptom Checklist-90-revision (SCL-90-R), were lower in individuals without PIU vis-a-vis PIU [16, 17].

Researchers have suggested various socio-demographic, personal, and internet-related factors are associated with problematic internet use; male gender [4, 18], initial years during the study course, the influence of peers, always logged in status, online interaction with friends, chat, watching porn, online new friendships or relationships, online shopping, average daily time spent on internet and internet access modalities were some of the risk factors for problematic internet use [2, 3, 18, 19]. However, correlates of problematic internet use among undergraduate medical students of North India remain underexplored. Moreover, no research studies have demonstrated a model predicting problematic internet use in a new subject. This study aims to assess the correlates of problematic internet use among undergraduate medical students of Delhi and derive a model for allocating new subjects among categories of internet users.

A cross-sectional study was conducted on 201 undergraduate students in a medical college of Delhi, India, from April 1st to May 31st, 2019. Inclusion criteria: All the undergraduate students studying in a medical college in Delhi and using the internet at least for the last six months were selected for the study. Exclusion criteria: those who do not give consent. Sample size calculation: A total of 300 students were enrolled in the medical college, 100 students in each batch. The sample size was calculated by taking prevalence of PIU as 50% at a level of 95% significance, 5% precision, and population size as 300.

n = [Np (1-p)]/ [(d²/Z²_1-α/2*(N-1) + p*(1-p)].

Where, n = sample size, N = population size, p = prevalence, d = precision.

The required minimum sample size comes up to be 169; Considering a non-response rate of 10%, the final minimum sample size was 188. We have studied and analyzed data from 201 students.

In the present study, we have analyzed data of 201 subjects; the majority of the study subjects were ≥ 20 years. Approximately 2/3rd of the subjects was males & 96% followed the Hindu religion. Majority of the subjects (76.1%) belongs to nuclear family and upper or upper-middle SES (84.6%). Approximately 3/4th of the study subjects (74.1%) had permanent residence in Delhi, and more than half of the subjects (55.7%) stayed in the hostel. 36.8% of the study subjects were in the third year, 36.3% in the second year, and 26.9% in the first year (Table 1).

Two third of the study subjects were started using the internet during their early adolescent period. Most of the study subjects (60.7%) used the internet for 6–10 years, and only 11.4% of study subjects used the internet for more than ten years. Almost all the study subjects had smartphones (99%), 51.7% had laptops, 31.3% had computers, and 24.4% had tablets. Only 15.4% of subjects had all the above electronic gadgets. Almost all the subjects (99%) preferred smartphones for internet access. The majority of subjects access the internet daily, more than half (51.7%) of the study subjects preferred night time to access the internet, and only 9% of subjects preferred morning time to access the internet. The majority of subjects (72.1%) used the internet less than 5 h a day, and only 6% used the internet for more than 10 h a day. More than half (50.7%) of the study subjects had spent less than INR 150 per month on the internet, and only 10% of study subjects had spent more than INR 500 per month on the internet. 60.7% of study subjects were permanently logged in, and 41.3% had PIU (Table 2).

Common reasons for internet usage found to be work or assignment, communication with friends, browsing, recreational or relaxation purposes, time pass, shopping, and news update, i.e., 99, 98.5, 98, 97.5, 96, 92.5, and 90% respectively. About 21% of the subjects used the internet for gambling (Fig. 1). Univariate analysis shows that internet use for emotional support, watching adult content, and gambling were significantly associated with PIU; however, in binary logistic regression, chatting, emotional support and watching online adult content were found to be significant risk factors for PIU (Table 3).

In Table 4, The test of equality of group means has been performed to measure each independent variable’s potential before the model is created. Each test displays the results of a one-way ANOVA for the independent variable using the grouping variable, i.e., Internet Users as the factor. If the p-value value is greater than 0.05, the variable probably does not contribute to the model. Wilks’ lambda is another measure of a variable’s potential. Smaller values indicate the variable is better at discriminating between groups. We have found strong statistical evidence of significant differences between means of average internet users and problematic internet users for only seven variables naming email time (in min), shopping time (in min), YouTube time (in min), WhatsApp time (in min), movie time (in min), download time (in min) and educational use time (in min). In contrast, insignificant variables are not suitable to discriminate between average internet users and problematic internet users (Table 4).

The step-wise discriminant analysis method has been applied for selecting the “best” variables to use in the model. The step-wise method starts with a model that doesn’t include any of the independent variables. At each step, the predictor with the largest F value to Enter a value that exceeds the entry criteria 3.84 is added to the model. At the last step, the variables left out of the analysis all have F to Enter values smaller than 3.84, so no more are added. So, the final selected variables in the model having F to enter value > 3.84 are family income, email time, and WhatsApp time. The F value for a variable indicates its statistical significance in the discrimination between groups, i.e., it is a measure of the extent to which a variable makes a unique contribution to the prediction of a group membership.

The equation of the model while considering the variables selected by applying step-wise discriminant analysis is as follows:

D = (0.000* family income) + (0.0076* email time) + (0.001*WhatsApp time)– 0.294.

We can calculate the discriminant scores by putting the values of these three variables in the above discriminant equation, by comparing this discriminant score with the cut-off value (Fig. 2), we can predict the allocation of subjects in average internet users or problematic internet users’ group.

Table 5 shows that 66.2% of respondents were correctly classified into average internet users and problematic internet users’ groups. This model correctly predicts 92.4% of subjects with average internet use.

Figure 2 shows the discriminant scores of 201 subjects. The centroid value for average internet users is − 0.278, whereas for problematic internet users is 0.395.

The general formula for the calculation of cut off value is given by

Where,

Z_CS = Optimal cut-off value between groups A and B.

N_A = number of observations in group A.

N_B = number of observations in group B.

Z_A = Centroid for group A.

Z_B = Centroid for group B.

So, in this case, the cut-off score will be 0.118. Hence the cut-off values above 0.118 are classified as problematic internet users, and below 0.118 are classified as average internet users.

In our study, we have found 41.3% of the subject had PIU; this has been corroborated with a study conducted on medical students by Pramanik et al. [9]; whereas in some other studies, the PIU ranges from 5.8–30% [3, 7, 8, 12, 23‐25]; however, a study conducted by Sayyah et al. [11] found a high prevalence of PIU (51%). The high magnitude of PIU in our study may be due to the demographic profile of study subjects as a majority (84.6%) belongs to upper or upper-middle SES and increased penetration of internet in metro cities like Delhi. We have found no significant association between gender and PIU; similar results were found in a study conducted by other researchers [26, 27], whereas most of the studies showed that male gender was significantly associated with PIU [2, 3, 6, 7, 23, 25, 28, 29] However, previous studies found that females were significantly associated with PIU [24, 30]. No significant association of gender with PIU in our study may be due to the fact that good accessibility to internet among male and female medical students. In this study, no association was seen between hostel accommodation and PIU; the same results were found by Salehi et al. [6] and Ghamari et al. [30]; whereas PIU was significantly higher in hostellers vis-a-vis non-hostellers in studies conducted by Chaudhari et al. [2] and Anand et al. [29]. We have found no association of PIU with the year of study, and the same result was found by Chaudhari et al. [2]. In contrast, Krishnamurthy et al. [12] and Asiri et al. [31] were found that students in first or second professional years had significant higher PIU as compare to third- and fourth-year students; however, Sayyah et al. [11] found that PIU was significantly higher in senior students as compared to junior students. The reason of no association of PIU with the year of study in present study may be due to sharing of similar psychological and environment factors among medical students of all the professional years. We have found no significant association of PIU with age at first use of the internet. In contrast, some authors found age at first use of the internet was significantly lower in students with PIU [2, 28]. We have found no association of PIU with a preferred time of internet use; this has been corroborated with a study conducted by Salehi et al. [6]; whereas Gedam et al. [3] found PIU was significantly higher in students whose preferred time of internet access was evening or night vis-a-vis morning or afternoon.

Our study shows that internet usage for emotional support, watching adult content, gambling and chatting was a statistically significant risk factor for PIU; this had been corroborated with studies conducted by several researchers [2, 7, 12, 23, 32]. We have found that internet usage for work or assignment, communication with a friend, browsing, recreational or relaxation, wasting time, shopping, news update, research, games, and meeting a new person on social media were not significantly associated with PIU, identical results were found in various studies conducted in India and other countries as well [3, 6, 7, 28]. In contrast, Salehi et al. [6] found that communication with friends was significantly associated with PIU & in a study conducted by Krishnamurthy et al. [12], internet use for work and making new friends on social media was significantly associated with PIU. A study conducted by Mazhari [7] found that the internet used for shopping was significantly associated with PIU. To the best of our knowledge, the present study is the first that discriminate a new subject in the average and problematic internet user groups. The model derived from step-wise DA suggests that family income, email time and WhatsApp time discriminate 66.2% of the subjects correctly into average and problematic internet user groups. WhatsApp is one of the commonest social networking applications used to share text massages, videos, photos, and work-related information which leads to its excessive use in everyday life [33]. Driving factor for the spread of WhatsApp use is its convenience; people may access massages and reply from anywhere anytime. Income is one of the factors that directly correlates with internet use. High income leads to more use of internet [34]. There is increase in the use of email for academic work, assignment and research related activity among medical students. These discriminators can be used to determine the PIU among undergraduate medical students.

Our study has several limitations; First, it’s a single-centre study, so multi-centre studies are warranted and explore the differences in areas, specialties, and grades. Second, being a cross-sectional study, we could not establish a cause-and-effect relationship; a longitudinal study would be more informative. Third, this study is subject to some recall bias. Fourth, the use of self-report for measuring time spent on a range of devices and activities by a person is likely to be biased.

Our study reported high PIU among undergraduate medical students. Internet usage for emotional support, watching online adult content, and chatting was significantly associated with PIU. We should create awareness among medical students regarding PIU and its potential harms; this could be included in the foundation course of the curriculum implementation support program (CISP) for MBBS students. The initiative should be taken to create ample opportunities for students to involve in extracurricular activities and interact with friends. There should be a provision of counsellors for emotional and mental support of medical students as they are overburden with studies and long posting schedules.