We conducted this study to assess the magnitude of school absenteeism and to study its correlates. Identification of social, demographic and medical correlates may help in predicting children at higher risk of absenting themselves and ensuring timely preventive interventions.

A cross-sectional, school based study was conducted in three government schools in South Delhi. The absenteeism was studied over a 6 month period from July to December 2006. Total of 704 children, of both sexes in the age group 10-15 years were registered in standards 6 to 9, in all the three schools. Each standard had 3 to 5 sections, varying across schools. Of the standards having 3 sections, one section per standard was randomly selected and all students in selected classes were eligible to be enrolled. Where there were more than 3 sections per standard, 2 were chosen randomly. Participants were included following an informed written consent. Repeat visits were undertaken to interview those who were absent at the first visit. Students who contributed only a few school days due to late admission in the current session or who left the school were excluded.

At enrolment, information on socio-demographic profile of the students was collected. It included age, sex, class, education and occupation of the parents, their family structure and income. The socio-economic status was calculated as per the Revised Kuppuswami’s Scale for determining socio-economic status of urban families (2001).

A pre-designed questionnaire was administered to ascertain the duration of absence and the causes for absenteeism, medical and non-medical. Participants were assured of confidentiality and were inquired about school truancy and various phobias of schools, teachers and subjects. The causes of absenteeism were also ascertained by school records, leave applications and one month’s recall by the students. Students, teachers and parents were interviewed whenever needed.

The average absenteeism of more than 2 days per month (i.e. 12 days in the 6 month study period), was defined as significant absenteeism, for the purpose of our study. Despite extensive review of existing literature, there is no consensus on the level of absenteeism which may be regarded as significant. Previous studies have considered absenteeism even when the child was absent on a single day, to define their own criterion for absenteeism(8-10).

Data were analyzed using Stata 9.1 software. The average number of days of absence in the 6 months reference for each child was calculated along with 95% confidence interval. The proportion of significant absentees was determined along with 95% confidence interval. The correlates of significant absenteeism were assessed by calculating the odds ratios. Stepwise multiple logistic regression analysis was performed to identify predictors of absenteeism.

A total of 704 students were registered of which 332 (47.16%) were boys. The mean number of days of absenteeism over the 6 month study period was 14.3±10.2 days (95% CI 13.5-15.0). The total number of working days was 140.2±8.6 days over the last 6 months. Hence, the average absenteeism per child was 10.2%. Only 9 children did not miss a single school day. Many had missed 1-6 days (26.6%), 6-12 days (24.4%), 13-18 days (17.1%), 19-24 days (10.2%) or 25+ days (20.4%). 336 (47.8%) children had significant absenteeism.

Table I



Relationship of Sociodemographic Factors with Significant School Absenteeism

Male sex, increasing birth border and family size, lower parental education and income were identified to be associated with significant school absenteeism (Table I). Causes responsible for their school absenteeism, as reported by the students are listed in Table II.

Table II



Causes of School Absenteeism Reported by Students

On stepwise multiple logistic regression analysis, gender, age group ,birth order, parents’ education and income, school phobia, school truancy, school load and absenting for family reasons were found to be independent significant factors responsible for school absenteeism (Table III). Based on these factors, we developed a model to predict absente-eism, taking the sum of the regression coefficients weighed by the code for each predictor. All the variable scores for a particular child were summed up to arrive at a final score.

Table III



Predictive Model Based on Multivariate Regression Analysis

We also found that school absenteeism had negative correlation with the academic performance of the students (r = –0.513).

The average absenteeism per child in our study is 10.2%. Gender (male sex) age group, birth order, parents’ education and income, school phobia, school truancy, school load and absenting for family reasons were found to be independent significant factors related to increased school absenteeism.

As compared to a study conducted by Awasthi, et al.(8) in 2000-2001 who calculated prevalence as 4.7%, the absenteeism has increased. However in New York(2), percentage of absenteeism varied between 7.3% to 17.8%. The factors found significant in our study are consistent with previous studies linking absenteeism to male gender(8), younger age(11), increasing birth order(12) and lower parental education and income(1,13,2). A different trend was seen in the NCHS study(2) where absenteeism was higher among older students. Ananthakrishnan, et al.(11) found no significant gender difference. The differences may be attributable to different settings of the study. Despite extensive research we could not find a multivariate analysis of the factors of school absenteeism.

We developed the model based on the factors found to be significant. Our prediction models appear to be useful for predicting prospective absentees incorporating relevant risk factors. There are no existing models to predict absenteeism. This model can be applied to all the students in the given setting; however, modification and further evaluation by receiver operating characteristic curve may be required when applied to a different setting(14,15).

Hence, school absenteeism has a high magnitude, with 48% children absenting themselves for more than two days per month. Our model predicts the chances of a particular child to be an absentee. The predictive value of the model is about 92.4% and can be used for timely preventive interventions

Contributors: PP conceptualized and designed the study, interpreted data, revised manuscript and approved the final version to be published. She will act as guarantor of the study. PU acquired the data, drafted the article and helped in final approval of the manuscript. VS analyzed the data, revised contents and helped in final approval.

Funding: None.

Competing interests: None stated.