Methods

In Scotland there are 190 days of schooling each year divided into three terms. We chose to define frequent medical absences as occurring when the student was recorded by the school as absent for “medical reasons” for more than 20% of possible attendances (an amount of absence equivalent to an average of one whole day per week) over three consecutive school terms. The absence was recorded cumulatively and frequent absence could be achieved either by continuous absence or by intermittent absence.

The study was conducted in two stages: First, we analysed existing routinely collected school survey data to determine the prevalence of frequent absence for “medical” reasons in the sample of schools and the characteristics of the absent students. Second, we conducted a case–control study in the same population to test hypotheses about the factors associated with such absence.

Survey

The survey was based on routinely collected school attendance data for the period August 2003 to October 2004 from the City of Edinburgh Education Department. Edinburgh is Scotland’s capital city and has a total population of approximately 450 000. Whereas Edinburgh is a relatively wealthy UK city in terms of average income per head, there is a wide range of income and 16% of secondary school students are eligible for free school meals. The survey data used were on 8839 students attending the first to fourth years in 10 of the 23 local authority secondary schools in Edinburgh. The schools were chosen to be representative of all those in the city in terms of size, socioeconomic status and religious affiliation. Only local authority schools were included because only they had a standardised method for recording school absence.

Case–control study

Cases and controls were identified from the same schools by prospectively monitoring the daily school attendance records. Cases were those students who accumulated absence exceeding the defined threshold. Students who had left the school during the study period were ineligible. Controls were selected from students with a good attendance record (in the lowest 10% of absence for their school year group) and were matched to the cases for age in years, gender and school class.

The student and their parents were contacted by a letter from the school, followed by a telephone call if they failed to respond. Written consent was subsequently obtained from the students and from the parents of those willing to participate. The research doctor (RJ) visited the cases and controls either at home or at school (depending on student preference) in order to carry out the research assessment.

The research assessment comprised the completion of questionnaires by the student and parents, an interview and a psychiatric and physical examination.

The questionnaires were a 17-item (16 for boys) check-list of common symptoms specifically designed for the study, which asked them to indicate which if any of the symptoms on the list “had bothered them a lot” over the previous 12 months (see appendix). The student and a parent or carer (giving their view of their child’s health) both completed the Strengths and Difficulties Questionnaire (SDQ),4 which assesses psychological adjustment. The researcher did not prompt the students with answers to questions but did respond to requests for clarification.

At interview the students were asked to give the reasons for their school absence, and information on their family composition. A full medical history was taken, which included questions to identify recurrent abdominal pain, chronic headache and chronic fatigue syndrome.

A physical examination was carried out by the research doctor (RJ) including general cardiovascular and respiratory examinations and a quick neurological screening test. The research doctor also administered a structured psychiatric interview, the Diagnostic Interview Schedule for Children (DISC),5 which yields psychiatric diagnoses compatible with the current International Classification of Diseases6 and the Diagnostic and Statistical Manual, version 47 classification systems.

Data on the occurrence and outcome of specialist assessments was obtained by a review of all medical and psychiatric records held on the student by local NHS hospital services. Social deprivation was assessed by postcode using the Depcat categories devised by Carstairs and Morris8 (seven categories, with 1 indicating lowest deprivation and 7 the highest).

Finally the medical and psychiatric diagnoses for the last 12 months were made by a second doctor who reviewed the data obtained from the research assessment and all available hospital records (medical and psychiatric). This second doctor was blind as to whether the student was a case or a control.

Ethical approval

The study was approved by the Local Research Ethics Committee and also by the Ethics Committee of the Education Department of City of Edinburgh Council.

Analysis

The survey data were analysed to determine the prevalence of frequent absence and associations with other variables (school, eligibility for free school meals and age) using the χ² statistic.

For the case–control study the sample size was calculated a priori on the basis that the base rate of psychiatric morbidity in the general population was 10% and that the minimum clinically important difference between cases and controls would be 20%. Consequently, a minimum of 82 subjects in each group would allow this difference to be detected at p<0.05 with 90% power. Conditional logistic regression was used to compare associations of hypothesised variables in cases and controls, and confidence limits for odds ratios (OR) were calculated using Wilson’s method. The use of specialist NHS services was described and the proportion of cases with psychiatric disorders who had not attended specialist services was calculated.

Results

Prevalence and reasons for absence

In the whole study sample of 10 schools and 8839 students, 198 (2.2%) met our criteria for frequent medical absence during the study period. The absence rate varied between schools and was greater in those receiving free school meals and in older students (all by χ², p<0.001).

Of the 198 potential cases, 46 had left the school during the study period. Ninety-two (61%) of the remaining 152 students were recruited to the study (three did not respond, 49 refused to participate and despite apparent agreement, interviews were not achieved in a further eight). Students who were not interviewed did not differ in age, gender or deprivation scores from those who participated. Ninety-two matched controls were also recruited from 136 eligible students approached (seven did not respond, 35 refused and interviews were not achieved on a further two despite apparent agreement to participate). The final sample was of 92 cases and 92 controls. Their characteristics are shown in table 1.

The main reason given at interview to explain absence was physical symptoms. The most commonly reported symptoms were aches and pains, asthma, cough, headaches, tiredness and abdominal pain; only four cases gave psychological problems as the explicit reason for absence. On the symptoms checklist the cases reported significantly more current symptoms than controls, although controls were not symptom free (see table 2).

Family composition

There was more family disturbance in the cases; the rate of divorce or separation was greater (53% vs 24%; p = 0.002) as was the percentage of students who were living with non-parental carers (17% vs 1%; p = 0.007).

Medical diagnoses

Only 8% (seven of 92) of cases and none of the controls had serious organic disease. The diagnoses were: asthma requiring hospital admission (two cases), cystic fibrosis (two cases), cholesteatoma with complications (one case), shunted hydrocephalus with complications (one case) and osteosarcoma (one case). Eleven per cent (10/92) of cases and no controls had defined symptom-defined syndromes. These were: chronic headache (one case), recurrent abdominal pain (two cases); chronic fatigue syndrome (seven cases). The remainder had only symptom complaints (such as pains in various sites) or illness of minor severity (such as mild asthma). Medical assessment did not lead to any new referrals for further medical investigation as all patients in whom this was considered were already under specialist medical care.

Psychiatric diagnoses

The psychiatric diagnoses made in cases and controls are shown in table 3. Forty-one out of 92 (45%) cases had had a psychiatric diagnosis during the previous year, compared with only 16/92 (17%) of controls (p = 0.001, 95% CI for OR 1.37 to 4.02). The commonest psychiatric diagnoses were emotional disorders (29 cases), comprising depressive disorders (16 cases), anxiety disorders (14 cases), obsessive-compulsive disorder (13 cases), post-traumatic stress disorder (five cases). Other diagnoses were oppositional defiant disorder (19 cases) and attention deficit hyperactivity disorder (13 cases). Co-morbidity (the simultaneous presence of more than one disorder) was common among the cases. The cases also had higher scores than the controls on all four subscales of the SDQ as rated by both the student and parent.

Hospital records and adequacy of treatment of psychiatric disorder

A review of the hospital records indicated that 38 cases and 10 controls had attended hospital for a specialist medical or psychiatric opinion in the study period. However, only 14/41 (34%) of the cases who had a psychiatric diagnosis had been seen by NHS specialist psychiatric services during the study period.

Discussion

This study has found that frequent medical absences from school (defined as missing more than 20% of possible attendances over three consecutive school terms for reasons given as “medical”) occurred in at least 2% of local authority secondary school students. Some students had accumulated their absence by being off for prolonged periods, others had very frequent short periods of absence. The association found in the survey between frequent medical absence and free school meals and being older probably reflects the effect of social deprivation and a greater tendency to be off school for all reasons with increasing age.

In the case–control study most of the cases gave physical symptoms as the reason for absence. On enquiry they also reported being bothered by more physical symptoms of all types than the controls (although it should be noted that the controls were not symptom free). In most cases, however, these symptoms were not associated with a serious organic disease diagnosis. A substantial minority of cases had symptom-based diagnoses. Chronic fatigue syndrome was the diagnosis in 8% of the cases. Although an important cause of medical absence, this figure is much lower than the 40% previously reported.9

The cases had a much higher rate of psychiatric disorder (mainly anxiety and depressive disorders) than the controls on student and parent-completed questionnaires and on a structured interview-based assessment. The 45% rate of psychiatric illness found in the cases was substantially higher than in the controls and much higher than the 11% reported by the recent UK Office of National Statistics report of mental health in children and adolescents.10

Only a minority of the cases with a psychiatric diagnosis had been referred to NHS specialist psychiatric services. The failure to refer for psychiatric assessment may have been because the student-reported physical symptoms were the main reason for absence. It is notable that many cases had also not had a specialist paediatric medical assessment. Although most cases reported attending general practice, the lack of specialist assessment is surprising given that these students had missed 20% of their schooling over the previous year for reasons described as medical.

There was also an association with family factors. The cases were more likely to have divorced or separated parents. This association may have many explanations; one might be that students are staying off to care for a parent who is physically or mentally ill.11 We were not, however, able to assess parental health in this study.

Overall, these findings are consistent with other studies of adults and children, which have found that physical symptoms and symptom syndromes are a common reason for incapacity and are often associated with psychiatric illness and personal and family problems.12 13 14 15

Limitations

The study has a number of limitations. First, we surveyed only local authority secondary schools in Edinburgh, UK. Consequently, our findings may not be representative of local authority schools elsewhere or of fee-paying schools. The schools included were, however, chosen to be representative of the city population and included students from a wide range of socioeconomic backgrounds. Second, in the case–control study we were only able to assess 60% of the eligible cases, meaning our sample may not be representative of the local school population. There were, however, no substantial demographic differences between the participants and non-participants. Third, it is possible that there was measurement bias in the assessments. This may have arisen because the research doctor was not blind as to whether the student was a case or control, although the final medical and psychiatric diagnoses were made by a second doctor who was blinded. Physical symptoms are common in the healthy adolescent population,16 and cases may have over-reported their symptoms in order to justify their absence from school. However, even if this had occurred it would not account for the low rate of diagnoses of organic disease found in the cases. Fourth, it is possible that the cases had undiagnosed medical conditions that were not apparent to us but which may have been diagnosed with further medical assessment and investigation. Although we cannot entirely exclude this possibility, we do not think it is likely to apply to many cases. All cases were examined by an experienced paediatrician and all cases were considered by a second doctor who also reviewed all available hospital case notes. Finally, we determined the unmet need for psychiatric and psychological assessment by whether or not the students had attended NHS child and adolescent psychiatry services in Edinburgh, and we did not record assessment in primary care and private care. However, given the very substantial school absence of these students, referral to the NHS psychiatric service would be expected if a psychiatric disorder was suspected.

Conclusions

To our knowledge, this is the first study of frequent absences for “medical” reasons among secondary school students. We found such absences to be relatively common. Importantly, only a minority of cases of frequent absence were associated with a diagnosis of serious organic disease or with a symptom-defined syndrome such as chronic fatigue syndrome. Rather, the main problems were medically unexplained physical symptoms and psychiatric illness. A substantial number of cases of psychiatric disorder had not received specialist assessment.

Further research might usefully explore the reasons for absence from school in more detail, perhaps using a qualitative approach. Studies of interventions aimed at helping students to achieve an earlier return to consistent school attendance are clearly required. Even before such work is done, the findings of this study should prompt education departments and their NHS partners to look more critically at the problem of “medical” absence from school and to establish a system that provides more comprehensive assessment and treatment. Such measures have the potential not only to minimise the disruption to their schooling but also to promote their long-term welfare.