-PEDIATRIC-
PREMIUM THESIS TOPICS

An epilepsy protocol is judged on internal consistency: the research question, objectives, methodology and statistical plan must all describe the same study. The primary objective should be a single measurable endpoint — one prevalence, one comparison, one association — with everything else demoted to secondary objectives. Beyond that, this subject turns on a handful of definitions, and each one has to be settled in writing before registration rather than argued at viva.

Name the classification and its edition, and decide in advance where the unclassifiable child goes. A page of focal-versus-generalised comparisons assumes every child can be placed in one box, and a real clinic cannot do that. State that the 2017 International League Against Epilepsy framework is being used, apply it at all three levels — seizure type, epilepsy type, syndrome where identifiable — and say explicitly what happens to the child who falls into "combined generalised and focal" or "unknown". Excluding them quietly inflates the clarity of the results; keeping them without a plan wrecks the analysis. Say who classified, on what information, and whether a second clinician reviewed the assignments.

Aetiology is limited by the unit's investigations, not by the child's disease. Report aetiology against the six standard categories — structural, genetic, infectious, metabolic, immune and unknown — and state plainly which investigations were available. A study in which magnetic resonance imaging is unfunded and genetic testing unavailable will produce a large "unknown" fraction that describes the hospital, and the protocol should say so rather than presenting it as a finding. Where computed tomography is what patients can afford, say that the imaging arm is CT-based, since a CT-detected calcified granuloma and an MRI-detected focal cortical dysplasia are not the same category of evidence.

Specify the electroencephalogram as a procedure, because the yield belongs to the recording. Diagnostic yield, interictal discharge frequency and background abnormality all depend on how the study was done. State the recording duration, the electrode montage, whether natural or sedated sleep was captured, whether sleep deprivation was used, and whether hyperventilation and photic stimulation were performed. State the interval between the last seizure and the recording, since discharges are commoner soon afterwards. State who reported it, their level of training, and whether the reporter had access to the clinical diagnosis — and for the concordance topic, that independent blinded reporting is required, with agreement reported as kappa. A yield study that reports one number without describing the recording is uninterpretable and unpublishable.

Define the imaging read the same way. State the field strength, whether an epilepsy protocol sequence set was used or a routine brain study, and who reported it. For neurocysticercosis, apply the accepted diagnostic criteria by name and record the lesion stage, because a single degenerating cyst and multiple calcified lesions behave differently. For perinatal brain injury, state how the perinatal event was ascertained, since retrospective birth history from a parent is not the same evidence as a documented record.

Exclude events that are not seizures, and say how. Breath-holding spells, syncope, tics, parasomnias, stereotypies, gastro-oesophageal reflux in the infant and psychogenic non-epileptic events all arrive in a paediatric clinic labelled as fits. Any study of epilepsy characteristics, drug resistance or medication burden must state the criteria used to confirm the diagnosis and the exclusion of non-epileptic events, and should record whether the diagnosis rested on witnessed description, home video or an ictal recording. This is also the reason a study of "uncontrolled seizures" that never questions the diagnosis will overstate drug resistance.

Name every instrument, with who administers it. Developmental, cognitive, behavioural, quality-of-life, adherence, stigma and caregiver burden topics all rest on a tool, and the tool is the study. State the instrument and version, the licence position, and the language, with forward and back translation documented where a validated local version does not exist. Developmental and cognitive testing must state who is qualified to administer it — a psychologist for a formal intelligence assessment, which a resident cannot perform alone — and must record whether the child was tested while on sedating medication, at what time of day, and how long after the last seizure, since post-ictal testing measures the wrong thing.

A paediatrics synopsis is the condensed document of two to four pages — title, introduction, aim and objectives, brief methodology, sample size and references — submitted for registration of the dissertation topic. The paediatrics protocol is the expanded version of twelve to twenty pages carrying the full review of literature, detailed methodology including case definitions, the electroencephalography and imaging procedures, the instruments with their scoring, the statistical plan, the study timeline and the annexures.

Four annexures carry particular weight on this subject. The case definition annexure should set out in one table how epilepsy, each seizure type, seizure freedom, uncontrolled seizures and drug resistance are defined for this study, with the source cited. The seizure record proforma matters more here than any other data sheet: a structured calendar or diary format capturing date, type, duration, time of day, precipitant and whether medical attention was sought, because a single question asking how many fits the child has had is the weakest link in most epilepsy dissertations. The instrument annexure reproduces each questionnaire with its scoring key and carries the permission or licence correspondence. And the consent set requires a parent or guardian information sheet and consent form plus a separate age-appropriate assent form in the local language.

Two paediatric specifics that examiners look for. The methodology must state the age range in completed years or months with both limits defined, and must say what happens to a child who ages out or transitions to adult care during the study. And where the study involves cognitive or psychological assessment, the protocol must state where the child is assessed, how long it takes, that the session does not displace clinical care, and what is done if the assessment distresses the child.

In practice the synopsis is extracted from the protocol rather than written separately, which is faster and produces a more coherent document. Check the university's prescribed proforma before submission, since rejections on formatting grounds are common and entirely avoidable.

Match the formula to the design. Prevalence topics — developmental delay, behavioural problems, drug resistance, vitamin D deficiency, seizure-related injury — use a single proportion formula with an expected prevalence taken from a cited comparable study, and where the finding is uncommon use relative rather than absolute precision. Comparative topics need a two-mean or two-proportion calculation with both expected values referenced. Correlation topics size on the expected coefficient. Diagnostic yield topics are sized on the expected yield with a stated precision. Name the citation that supplied the input; an unreferenced expected prevalence is the commonest reason a calculation is returned.

Seizure counts are not normally distributed and should not be analysed as though they were. The distribution is skewed, bounded at zero, and dominated by a minority of children with very high counts, so a mean seizure frequency with a standard deviation describes nobody. Summarise as median with interquartile range, compare with non-parametric tests, or model the count directly with Poisson or negative binomial regression, offset by the observation period. Where the analysis needs a binary outcome, define it in advance — seizure freedom over a stated interval, or a threshold number of seizures — rather than splitting at the sample median after the data are in, which is not reproducible and cannot be compared with anything.

Prevalence measured in a paediatric neurology clinic is a referral figure, not a population figure. Drug-resistant epilepsy will look far commoner in a tertiary unit than in the community because resistant children are precisely the ones referred and retained, and the same filter inflates structural aetiology, polytherapy and developmental delay. State this as a sampling limitation in the protocol rather than in a single line of the discussion, and word the objective as prevalence among children attending the clinic. Prevalent-case sampling has a second consequence that affects several topics here: children with long-standing severe epilepsy accumulate in a clinic population, so duration of epilepsy and poor cognition will correlate for reasons that have nothing to do with duration causing anything.

Watch for common-informant bias in the psychosocial group. Child behaviour, quality of life, adherence, stigma and caregiver burden are frequently all reported by the same parent in the same sitting, and a distressed caregiver rates the child worse on everything. The correlation that results is partly real and partly an artefact of a shared reporter. Mitigate it where possible — a teacher-completed behaviour form, child self-report for the older child, an objective adherence measure such as pill count or refill records alongside the questionnaire — and where mitigation is not possible, state the bias explicitly and do not describe the association as an effect. Note also that parent-proxy and child self-report quality-of-life scores diverge systematically and must not be pooled into one mean.

The vitamin D comparison needs season in the model. Baseline deficiency in Indian children is high enough, and seasonal variation wide enough, that a comparison of enzyme-inducing against non-enzyme-inducing medication conducted across different months can produce a difference that reflects the calendar. Record the month of sampling, an estimate of daily sun exposure, skin covering, supplementation and dietary calcium, and adjust for them. State the assay and its cut-offs, since deficiency thresholds differ between guidelines and the reported prevalence moves with the threshold chosen.

Plan for confounding and for multiplicity. These are cross-sectional studies, so an association is not a sequence: polytherapy and poor cognition are measured on the same day and the child with the worse epilepsy was given more drugs. Word objectives as association, and adjust by multivariable regression for the confounders that matter here — age, age at onset, epilepsy duration, aetiology, seizure frequency, number of medications and socio-economic status — entered because they are clinically justified rather than because they survived univariate screening. Where the analysis carries a battery of subscales or a panel of tests, nominate the primary endpoint in advance and treat the rest as exploratory, or apply a stated correction.

Name the tests. Continuous variables are summarised as mean with standard deviation where normally distributed and median with interquartile range otherwise, with normality formally tested. Two independent groups use the t-test or Mann-Whitney U test, three or more groups analysis of variance or the Kruskal-Wallis test with a stated post-hoc correction. Proportions use the chi-squared test with Fisher's exact test for sparse cells. Agreement, including electroencephalography and imaging concordance, is reported as Cohen's kappa with its interpretation, not as a percentage of cases matching. For record-based components, state in advance how missing data are handled and report how many records were excluded for incompleteness.

Start from the clinic register. Count how many children with confirmed epilepsy attend in a year, how many are new versus follow-up, and how many fall in the age band the topic needs. A topic requiring eighty school-aged children with epilepsy is straightforward where a dedicated paediatric neurology clinic runs weekly and difficult where epilepsy is seen within the general out-patient department alongside everything else.

Then check what the unit can actually deliver. Electroencephalography availability, waiting time and whether sedation for a young child is possible decide every EEG topic. Magnetic resonance imaging that patients must fund themselves decides every imaging topic, and if only those who can pay are scanned, the imaging series is a socio-economic sample. Formal cognitive assessment requires a clinical psychologist and a licensed test, and if neither is available the topic cannot be done however good it looks on paper. Topics built on clinical assessment, a structured seizure record and a validated questionnaire are the ones that finish on time.

Descriptive clinical, aetiological and electroencephalographic profiles remain the most common and are readily accepted: a defined group of children with epilepsy, described across clinical, investigative and developmental parameters. Prevalence studies of a complication or comorbidity, comparative studies between epilepsy types or against healthy controls, and analytical studies associating an investigation finding with seizure control are all established.

Diagnostic yield and concordance studies form their own category and publish well, provided the reading is blinded and agreement is reported properly. Questionnaire-based studies of quality of life, adherence, stigma, caregiver knowledge and caregiver burden are accepted and are often the most feasible option where investigation support is limited. Prospective observational follow-up over the thesis period is possible for adverse drug reactions and seizure recurrence, but the cross-sectional framing on this list is deliberate, because it fits the interval between ethics clearance and submission.

Bring three to five shortlisted titles rather than one, since guides frequently rule out a topic on grounds a new resident cannot see — a senior resident already working the epilepsy clinic, a departmental project on the same register, an EEG machine due for replacement.

Settle six things in that meeting: how many eligible children attend annually and how the denominator was counted; who will classify the epilepsy and whether a second clinician will verify it; who reports the electroencephalograms and whether blinded reporting is possible for a concordance study; whether imaging is funded and by whom; which instrument will be used for development, cognition, behaviour or quality of life, whether a licence and a validated local-language version exist, and who is qualified to administer it; and which journal the eventual paper is aimed at. Where a topic needs psychology, radiology or the school system, secure that cooperation formally rather than on an informal understanding.

Some can, and record-based work is often the only way to assemble adequate numbers of an uncommon presentation within a thesis period. State the archive period, how records were identified, and the exclusion criteria, and apply for a waiver of consent explicitly rather than assuming one follows from the design.

Three cautions specific to epilepsy. Seizure frequency is rarely recorded well enough to analyse. Case notes carry impressions such as improved or occasional fits far more often than counts, so a retrospective study of seizure control usually ends up recoding vague phrases into categories, which is unreliable and should be declared if done. Prospective collection with a structured seizure record is better and costs only time. Classification in old notes reflects old terminology. Records written before the current framework use terms that do not map cleanly onto it, and a retrospective classification study is partly a study of how the notes were written. State how legacy terms were translated and by whom. Adherence, behaviour and quality of life are not in the record at all. Those topics have to be prospective.

The synopsis is the condensed two to four page document submitted for topic registration. The protocol is the full document of twelve to twenty pages containing the detailed review of literature, methodology with case definitions and the electroencephalography and imaging procedures, the instruments and their scoring, the statistical plan, the timeline and the annexures including the seizure record proforma and the consent and assent set. The synopsis is normally extracted from the completed protocol.

Institutional ethics committee approval before any data collection, under the national ethical guidelines for biomedical research involving human participants and the specific provisions governing research in children.

Consent and assent. Written informed consent from a parent or legal guardian, and written assent from the child from about seven years of age, in an age-appropriate document in the local language. State that participation is voluntary, that refusal does not affect the child's treatment, and that the child may decline even where the parent has consented. For an adolescent completing a questionnaire about mood, stigma or social difficulty, state who else sees the responses, because a parent sitting alongside changes the answers and may create a problem at home.

No change to treatment and no investigation for research alone. The protocol must state explicitly that no medication is withheld, reduced or altered for the purposes of the study, and that no child is imaged, sedated or sleep-deprived for research. Only electroencephalograms and scans already indicated clinically may be used. Where an electroencephalogram is being done anyway and the study adds an activation procedure or a sleep recording, that addition needs its own justification and its own line in the consent form.

Incidental and predictable findings need a named pathway. Cognitive and developmental assessment will identify previously unrecognised intellectual disability or learning difficulty; behavioural screening will identify attention-deficit hyperactivity disorder and, in adolescents, depression and suicidal ideation; imaging will occasionally show something unexpected. The protocol must state in advance who informs the family, how, and how the child is referred for assessment and support — not that findings will be reported in the thesis. A study screening adolescents for emotional difficulty without a counselling pathway should not be approved, and committees increasingly say so.

Video, stigma and confidentiality. Seizure video is identifiable by definition and requires separate written consent specifying who may view it, where it is stored and for how long; identifiable facial images should be avoided in any publication. Stigma and misconception questionnaires can distress a caregiver who recognises their own beliefs in the items, so state that the interviewer is trained, that questions may be declined, and that accurate information is offered afterwards. Where school data are collected, state that the school is not informed of the diagnosis without explicit family permission, since disclosure carries real consequences for a child with epilepsy in India.

Clearance commonly takes six to ten weeks and retrospective approval is not granted.

This is the weakest point in most paediatric epilepsy dissertations, and it is the one an examiner goes to first, because around thirty topics on this page use controlled, uncontrolled, seizure frequency or drug-resistant as an exposure or an outcome. Two separate problems have to be solved in the protocol. First, seizure counts come from a parent's memory, and memory is worse for some seizure types than others. A generalised tonic-clonic seizure is remembered; a brief absence, a nocturnal event, a subtle focal seizure with retained awareness and a single myoclonic jerk are not. So misclassification is not random, it is differential by seizure type, and the child with absence epilepsy can appear beautifully controlled precisely because their seizures are the ones nobody witnesses. Fix a defined observation window — commonly the preceding six or twelve months — state it in the objective, anchor recall to something the family remembers such as a festival, a school term or an admission, and use a structured calendar rather than a single question. Where the study period allows, issue a seizure diary at recruitment and collect it at review; a prospectively kept diary is worth more than any amount of careful questioning afterwards. Say who reported, since an adolescent and their parent give different counts. Second, drug resistance has a formal definition and it is not simply still having seizures on treatment. The accepted definition requires failure of two tolerated, appropriately chosen and adequately used antiseizure medication schedules, whether as monotherapy or in combination, to achieve sustained seizure freedom. Every clause does work. Appropriately chosen excludes the child made worse by carbamazepine given for a generalised syndrome. Adequately used excludes the child never titrated beyond a starting dose. Tolerated excludes the drug stopped for a rash in the first week, which was never a trial. And the whole definition presupposes the diagnosis is correct, which excludes the child whose events were never epileptic. A study that labels children drug-resistant without checking dose adequacy, adherence and diagnosis is measuring pseudoresistance and will report a prevalence far above the true figure. Build the checks into the proforma: current drugs with doses in milligrams per kilogram, duration at that dose, reasons for any drug stopped, documented gaps in supply, and an adherence measure. Report how many children screened as uncontrolled were reclassified as pseudoresistant — that number is frequently the most interesting finding in the thesis, and it is publishable in its own right.

Substantially. A PhD proposal typically runs 6,000 to 10,000 words and carries an extended critical literature review, a theoretical framework, a detailed methodology chapter and a discussion of expected contribution to the field. An MD synopsis is a two to four page registration document. The research question can be the same; the depth expected is not.

Most universities require registration within six to nine months of joining. Shortlist in the first two months, finalise with the guide by the third, and file for ethics clearance immediately afterwards. Recruitment here is slower than residents expect, because a child must first have a confirmed diagnosis, cognitive and questionnaire sessions take far longer than a case record entry, and any design using a seizure diary needs the diary period to elapse before the child can be analysed. Close the data collection window at least six months before submission.