Small and Special: the Hospital for Sick Children at Great Ormond Street Historic Patient Admissions Database
London, Kingston University, 2007
Churchill College, University of Cambridge
Date accessed: 16 April, 2014
Small and Special is the database of the historic admission records for the Great Ormond Street Hospital for Sick Children. The database contains information relating to all 84,190 individual patient admissions from the day the hospital opened in February 1852 until the end of 1914, and a further 10,290 admissions to the hospital’s convalescent home, Cromwell House in Highgate, covering the period 1869 when it opened, until 1904. The name is taken from Elizabeth Lomax’s book, Small and Special: the Development of Hospitals for Children in Victorian Times: at approaching one million records the database can hardly be called small, but it has every claim to be special.
The first dedicated children’s hospital in Britain, the hospital did start small, with only ten beds, soon rising to 20. It expanded to 75 beds in 1858, 120 years later, and increased capacity yet again in 1893. In the early days the hospital admitted only a few hundred children a year, but by the early 20th century there were up to 3,000 in-patient admissions annually, plus over 20,000 out-patients. The latter, however, are not included in this database. A substantial number of children were transferred to the convalescent home of Cromwell House, one of a range of convalescent institutions used by the hospital, which opened with 52 beds in 1869, rising to 150 by 1885, and which catered mainly for the youngest convalescent patients who would not be admitted to private institutions, and those with chronic complaints. The website mediates access to a machine-readable database of the transcribed inpatient admissions registers, covering children aged from a few days old to 16 (although the oldest patient was actually aged 30, she was admitted only because she was a nurse at the hospital who had fallen ill with scarlet fever). For children under the care of the hospital’s key founder, Dr Charles West, the admission record is linked to a scanned copy of Dr West’s case notes. These are supplemented by a library containing a series of articles on the founding of the hospital, the history of its buildings, departments, and associated convalescent homes, and the daily life of the children in the hospital and their parents. There are also brief biographies of 21 doctors, a handful of nurses, a dozen patients, the founders and other influential supporters and benefactors such as Charles Dickens, and these are complemented by a gallery of photographs and contemporary illustrations.
The admissions registers themselves consist of one record for each admission, each record containing the child’s name, age, gender, address, the name of the admitting doctor, the illness the child was suffering from, the dates of admission and discharge, where they were discharged to, and the outcome of the treatment. All these have been entered exactly as given in the original documents, and in addition other fields have been added: standardized names, addresses and diseases providing enhanced search capacity, and other derived fields generating useful information from the existing data, such as length of stay and year of birth. The website contains clear and relevant information about the generation of the database, the data entry and checking procedure, which fields are added and how they have been derived. Records relating to children admitted less than 100 years ago have been anonymised by hiding the name data, but with great efficiency, ‘de-anonymisation’ is taking place on a daily basis as the 100 year mark is passed. Thus on 9 June 2009, it is possible to see the names of those children admitted on 9 June 1909, but not those admitted the following day. Data entry was performed by volunteers, but was painstakingly double checked to ensure accuracy: the process being explained in detail in the section ‘about the project’, where the various individuals involved in the project are also listed. The supporting articles are interesting, appropriate, scholarly and answer many questions about the history, development and operation of the hospital.
The website provides a simple search facility enabling searching by name (standardized forename and non-standardized surname) and year of birth, with an option to specify various bands of years either side of the specified date. A more comprehensive search facility permits more flexibility in searching for names starting with a given letter or group of letters, and also sex and age on admission (within a chosen range). While it is not compulsory to register with the database, only these basic search facilities are available for non-registered users. Registration, which is simple and painless, opens up the possibility of searching on virtually every field in the database, neatly arranged in four panels: personal details (names, sex, year of birth, age on admission), residence (location keyword, street name, registration district and registration sub-district), admission and stay (hospital admitted to, date of admission, admitting doctor, length of stay in days, and whether case notes are available), and disease, outcome and discharge (disease name, disease group, ICD10, outcome, where the child was discharged to, and date of discharge). Searching is aided by drop-down lists of the possibilities where these are not too numerous and, for dates and ages, the option to specify either a precise date or age, or a range. Incredibly detailed and precise searches are therefore possible, and research is aided by the facility for registered users to download sets of complete records, which can be used for subsequent analysis, although these are limited to only 200 records per search. Contact with the website team is possible using a web form, and it is also easy to report errors or discrepancies with particular records in a similar fashion.
To get and download a few fairly random samples of records, but which would also preserve the structure of the data by allowing for some clustering of individual children, I searched for some Christian names (e.g. Agnes, Rachel, Francis and Frances) and for some surnames (e.g. surnames beginning with Ve), and downloaded the resulting datasets. To generate figures for the mortality rates and numbers in this review, I searched for particular year, age, disease, and outcome combinations, and noted the numbers of records returned by the searches to use as denominators and numerators for my calculations. I also searched for interesting records by selecting ‘absconded’ under outcome, lengths of stay of over a year, and other extreme results.
The database can yield gems of poignant information about individual children, such as the nine-year-old Edward Mason of Soho, first admitted in the spring of 1885 with enteric catarrh and tuberculosis, leaving (pronounced cured) after nearly three months in first Great Ormond Street then Cromwell House. Early the following year he returned, suffering from bronchitis and bronchopneumonia, to stay another three months between the two hospitals. He was admitted yet again in the spring of 1887, with heart disease and phthisis (pulmonary tuberculosis). It clear that after two months confined in Great Ormond Street for the third year in a row, and facing the prospect of another month or so in Cromwell House, the 11-year-old had had enough of institutions, and despite his ill-health, ran away after only two days at the convalescent home. Cases such as this – and there are many such examples of children going in and out of hospital for often quite lengthy periods – flesh out images of sickly Victorian children, dogged by chronic illness, as epitomized by Dickens’ Tiny Tim. Case study type examples on the website in the form of biographies of patients (which can be found in the library section, a rolling example also always displayed on the home page) show how individuals and their families can be traced in censuses and other sources, illustrating clearly how the admissions registers can provide additional details about children’s lives, invaluable to the family historian, genealogist or social historian wanting particular examples.
However, as well as illustrative examples, and the linking of individuals to other sources, the database can also yield more statistically robust data, such as mortality rates according to age, time period, or cause of death. For example I was able to calculate that as expected, mortality rates declined with age from 350 deaths per 1,000 admissions of infants, to 36 per 1,000 admissions of 10-year-olds (I did not perform calculations for ages above 10). Mortality also declined over time: looking just at those aged one year on admission, mortality declined from 316 per 1,000 in 1862–71 to 225 per 1,000 in 1902–11. Over half of all one-year-olds admitted with tubercular diseases died, and nearly four in ten of those in the same age group admitted with infectious fevers, but the pattern was different for older children: one in ten, or 100 per 1,000 10-year-olds admitted with tubercular disease died, but the rate was less than half that (43 per 1,000) for 10-year-olds admitted with infectious fevers. These brief calculations raise a plethora of interesting questions which could be examined using this data, such as the extent to which the changing mortality in the hospital was a reflection of the changing balance of diseases and conditions admitted, the age structure of the patients, and the virulence and treatment of different diseases. Also important is the issue of how repeated spells in hospital by the same child should be treated: many children endured repeated, non-contiguous spells in hospital, with different outcomes: if a child is ‘relieved’ or ‘not-relieved’, rather than ‘cured’, should the next spell be treated as part of the same disease episode or not? The answer might be different for different diseases: children with cleft palate and hare lip appear to have had repeated visits at which the outcome was reported to have been ‘relieved’ before a final ‘cure’ was achieved. Episodes of infectious diseases such as scarlet fever, on the other hand, were unlikely to have occurred more than once to the same child. Comparisons of length of stay could be thrown into the mix and many more questions can be asked, and the database is large and diverse enough to provide a mine of information for rigorous academic research, from the operation of and access to hospitals, in depth investigation of morbidity and mortality, and the impact of medical advances and disease environments.
Despite an attractive, easy to navigate and informative website, there a few areas where some improvements could be made. It is a shame, for instance, that there is no scanned example page of an admission register to enable us to envisage how they looked in the originals. Dr West’s case notes are fascinating, but although numerous at over 1,300 admissions, they do not provide a representative sample of children or diseases, in terms of either the time period or age of child. In terms of time period they are restricted to the first two decades when Dr West as active in the hospital, and in terms of age, none of his case notes relate to infants, as he was opposed to the admission of under one-year-olds and his influence was such that the hospital admitted very few until his resignation in 1875. The balance could be rectified if more money were sought to scan and add case notes from other doctors (some of which clearly exist given some of the patient case histories provided), not only to provide examples from a different doctor who may have taken a different approach or specialized in different diseases, but also covered a different era and a different age range of children.
Most of the data is transcribed exactly as it appeared in the registers, including, therefore, inconsistencies and earlier transcription errors, for example between ward books and admission registers. Some of these have been ironed out by the addition of standard fields, but some remain, such as the significant number of children with unexpected forename and sex combinations (where the case notes exist these could be ironed out, but have not always been). Forenames with no clear gender assignation (such as Frances and Francis) may have been particularly prone to mistakes as there are several repeat visits by children who appear to change sex between visits, and it is not clear whether these were mistakes in the original data, or in recent transcription. In an ideal world these instances could have been flagged and corrected by recourse to other records if possible (eg census records, other admissions, case notes etc), but such an exercise can be time-consuming and expensive in return for only small increases in accuracy, and it is possible that resources for this were not available. Despite the double-checking described, moreover, some clear transcription errors have not been picked up, such as the case where the disease was transcribed into the admitting doctor field, resulting in one of the doctors in the drop-down list possessing the unusual name of ‘Psoas Abscess’.
Not all fields amenable to standardization have been standardized: admitting doctor and surname are notable omissions, for which standardization would have considerably aided searching for both groups of and individual children. Care must also be taken when using standardized and other derived fields, as there is a certain lack of completeness and consistency. The records for Cromwell House, added later, do not have the standardized fields ‘residence street’ and ‘ICD10 codes’ or the calculated field ‘length of stay’ filled in: thus searching and sorting by these fields can be misleading as not all records fulfilling the search criteria will be selected. Standardization of names, places and diseases is a painstaking and difficult task, and it is easy to make mistakes and leave inconsistencies in the standard dictionary, but these are diminished when a dictionary is created from the finished data set rather than standardization performed as the data is entered. Some of the fairly obvious inconsistencies in this data suggest that standardization occurred in the latter manner, and they also prompt worries about the possible existence of less obvious mistakes. For example, there are cases in which streets have been coded to different districts or sub-districts: e.g. the address ‘55 Lilfud Rd, Camberwell’ is coded to residence street: Lilfud Road, Registration district: ‘Camberwell’, and the practically identical address ‘55 Lilford Rd, Camberwell’ is coded to residence street: blank and registration district: ‘Outside London’. Similarly the several variations of ‘Bengworth Rd, Loughborough Junction’ (Bangworth, Bensworth, Blagworth, Bengemont, Bougworth, Bengeworth) have been variously coded to the Registration District of ‘Lambeth’, with residence street ‘Bengworth Road’, ‘Blagworth Road’ or blank, or Registration district ‘St Saviours Southwark’ and residence street blank. Similarly, the ‘major standardization effort’ applied to diseases, has not resulted in an entirely standard classification: for example, a search for ‘infantile paralysis’ under disease, returns 549 records, while as search for ‘paralysis infantile’ returns an entirely different 109 records.
A brief definition of disease groups is provided in the help/diseases section, but these categories are fairly broad and each contains many individual diseases: it would also be beneficial for non-specialists in the history of infant diseases to have had a glossary of more specific illnesses.
Where a child has been referred from GOSH to the convalescent home of Cromwell House, a linked field is shown, enabling continuous stays in the related institutions to be easily viewed. However the claim that therefore ‘a user can view all records related to a child’s admission to the whole institution, following its progress from hospital to convalescent home, and sometimes back again’ (general help page), is misleading, as non-contiguous return visits by the same child have not been linked, but this information is only tucked away on the ‘hospital records help’ page. It is important to remember, therefore, that these are different admissions, not necessarily different children, or even different disease episodes, and therefore do not provide the whole story about each child. While it is possible to extract and link the data relating to different admissions by the same children oneself, it is perhaps one aspect of data enhancement which would have added considerable value to the data set, enabling not only the ability to follow a more complete and longer trajectory of a child’s illness or illnesses, but also a considerably more nuanced view of morbidity, case fatality and treatment procedures.
Finally, I found the restriction to 200 downloadable records frustrating, as it was frequently fewer than those found by my search criteria. Ingenuity in searching can enable the extraction of contiguous datasets which can be added together (for example by limiting the searches to a small number of years, children whose surnames begin with a particular letter etc, and repeating the searches by varying forename, surname or year) but this can be time consuming and cumbersome, and those who wish to do analysis on relatively large datasets might be best applying to the project managers for less restricted access to the data.
Most of these minor criticisms can be easily rectified with a little attention, such as an image of an admission register, tidying up the dictionaries, and extending calculated fields to the Cromwell House admissions. Others – such as adding more case notes from other doctors, linking individual children, would require more time and money, but would be well worth doing if resources allowed. Nevertheless, this is an excellent website, providing general access to a database which would be otherwise neither easily accessible and nor searchable, and which has been considerably and thoughtfully enhanced by additional fields and explanations. The detail and flexibility in searching, together with the contextual information and support, make it enjoyable to use and render it accessible and attractive to a wide range of users, from family historians to a variety of academics.
We are very grateful to Dr Reid for her detailed, thoughtful and largely positive review of the website. The team at Kingston has worked hard to produce what we hoped would be a useful resource for many different types of enquirers, and, perhaps, to show others what might be done with types of institutional records. This response deals with the detailed criticisms and gives some background to the project.
What is now the Historic Hospitals’ Admission Registers Project began at the beginning of the century, when it was envisaged that the end product would be a compact disk containing an enhanced database of the information contained in the hospital admission registers of Great Ormond Street Hospital (GOSH), from 1852 until 1903. As with almost all projects, the methodology took time to be perfected, and challenges – both technical and practical – presented themselves at every turn. The demand by researchers for data covering the crucial Edwardian and early Georgian period encouraged us to extend the original project to 1914. By this time, technological advances demanded a website, which was never part of the original plan. Demand created its own supply; and, with each enhancement of the project, more was requested by users. We were (and are) committed to a free site, which brings its own challenges, as well as rewards.
It is heartening to see the use that an imaginative and experienced demographer could make of even a limited number of downloaded records. Most of the shortcomings highlighted by Dr Reid have already been identified by the team, and stem from the fact that the project is one that, like Topsy, ‘just growed and growed’. If we knew where we were going to end up, some of us might never have started, but we are all agreed that we would have asked for far greater resources from the outset. If I may, I shall deal with Dr. Reid’s points as they arise in the review. The size of the databases, alas, is not one million records (if only), but nearer 100,000. At present, the site has the admission registers for the main GOSH and those of the Hospital Convalescent Home. Records for the Evelina Hospital in Southwark, the Alexandra Hip Hospital in Queen Square, London, and Yorkhill Children’s Hospital, Glasgow, will be added during the next 12 months. We would dearly love to add other hospitals (paediatric and otherwise) to the databases, and have plans so to do, but this depends on institutional support and future funding. Out-patients do not feature in the database (except when they are admitted to the convalescent home), as only one early out-patients register survives for GOSH. This is a great pity, as we are conscious that the vast majority of patients never saw the wards, but a trawl of the archive has only unearthed this lone volume for GOSH and the team is not aware of the survival of outpatient records for any of the other children’s hospitals.
We would be happy to add images of the admission registers, and shall talk to the GOSH archivist to arrange to have this done.
The inclusion of the case notes for Dr Charles West was a late development. The team was aware (as are other researchers in medical records) of the richness and potential value of the information contained in case notes. We are also acutely aware that these records are not protected under the Public Records Act. Adding Dr West’s case notes represented not only a technological problem to be solved, but was intended to alert repositories and other projects to the potential for unlocking the wealth of data contained in the surviving volumes. The process was time-consuming and involved volunteer indexers and medical historians, as well as paid conservationists, and IT specialists. Since it was undertaken, we have been looking at ways in which more case notes might be added, economically and efficiently. To that end, we have identified two physicians at the hospital, whose case notes span the 1870s to the early 1900s, that we believe would add immeasurably to the usefulness of the website. We are applying to the Wellcome Trust for the money to undertake this, on the understanding that the project entails advanced technological developments that have yet to be tried and tested in the field. These two doctors do have very young patients, although it is becoming apparent that GOSH’s enduring policy of restricting the under-twos on the wards meant that there were far fewer babies and toddlers than appear, for example, in the Evelina and at Yorkhill.
The points regarding the criticisms of data reliability are more difficult to answer. At the beginning of the project, it was decided that, in order to respect the integrity of the original documentation, the data were to be transcribed as seen. Mistakes regarding forename and sex entered by the original hospital staff would be left in the original field, but would, it was hoped, be corrected in the standardised fields. A validation programme and process was created to spot and rectify obvious errors, and we will run this again, to pick up mistakes regarding forename and sex that were missed first time around.
The point about the doctors’ surnames is well taken. The mistakes crept in after Cromwell House was added to the website, and we failed to pick them up. This suggests that we were not as careful in our checking of the convalescent home records as we should have been, and this will be rectified. The gaps in the Cromwell House data are entirely due to lack of money, and, as a consequence, lack of time. We had to stop the standardisation of the street names in order to get the data on the site on time, and had no funding to apply ICD10 codes to this set of data. It is a great pity, and makes the site less useful than it might be. We have not given up hope that we might be able to complete this, but it is just not feasible at present.
We are acutely aware of the defects on the site regarding Cromwell House, and our mitigation is that the work was done on a shoestring, and we decided it was better to have something – no matter how flawed – on line, than nothing at all.
With regards the other point re the admitting doctor; we plead not guilty. The names of the admitting doctors have been standardised, and included in the search form as a drop-down under ‘Admission and Stay’.
The lack of surname standardisation is mainly due to lack of time and resources; the decision not to use soundex was taken after much debate. Who is to say what is the ‘correct’ spelling of a surname, and would we have been justified in imposing standardised spellings, when even the most common surnames have alternative forms?
Dr Reid’s view on the standardisation process is perhaps open to debate. The standardisation was undertaken towards the end of the process, prior to the application of metropolitan registration and sub registration districts. While it is possible that we might have been able to undertake the standardisation of addresses at the same time as inputting the data, there is a limit to the number of fields and the amount of data you can reasonably ask a volunteer to fill in at any one time. The additional fields would, in our opinion, have been putting too much of a burden on the volunteers, and, thus, the standardisation was done retrospectively, by one member of staff. Attributing registration and sub registration districts was immensely complicated and time-consuming, with consistency being a persistent problem; there are very many High Streets and Church Roads in the registers, and, ultimately, educated guesses had to be made in some instances. Nevertheless, we are committed to re-checking the data, and to making changes when we are alerted to errors and mis-attribution.
The disease data could not have been standardised as we entered the data. It was imperative that we gained a ‘feel’ for the sorts of disease descriptions that were being used, and this knowledge only came with time. Disease standardisation took not months, but years, to formulate to our satisfaction. We held meetings with medical historians and paediatricians, and put each stage of our thinking through a process of scrutiny before key decisions were made. The realities of using volunteer in-putters precluded any other approach, and we felt we could not ask them to apply fairly complicated standardisation techniques while they were in-putting the data. In future, it may be that a technological means of achieving the simultaneous application of standardisation might be found, but one works with what one has, and within very real time and resources restraints. The errors in standardisation highlighted by Dr Reid are very regrettable, and, once more, I plead lack of sufficient technical support in creating the validation programmes.
The lack of a glossary of medical terms is deliberate, as we decided that it would be reinventing the wheel, and our limited resources would be better spent elsewhere. Accordingly, we have provided what we hope is a comprehensive list of sites which have this information. It is debatable whether we could have put these links on the disease fields, but, at present, they are to be found on the Links Page. It would have been rather wonderful if we could have developed a disease classification with narrower groups, but time was not on our side, and this was one decision that was dictated by the realities of deadlines. We did talk at length about linking multiple admissions for one patient, but ran out of time before this could be done. The statement on this in General Help is misleading, and reflects our original intention, rather than the current reality, and this will be amended accordingly.
The most serious issue for researchers is, we know only too well, the restriction on the number of downloaded records. This has very real implications for academic researchers, and we have been – and are – as accommodating as possible to those who need to download large numbers of records. The reason for the restriction is to protect the work of many people – much of it unpaid – from hackers and those who might want to download the whole database and make illegal use of it. We will always treat requests for access to more material sympathetically, but this project has taken years to get to this stage, and we have to be aware of the unscrupulous who might wish to profit from that. If anyone else has a solution to this dilemma, we should be very happy to hear it.
It is gratifying that Dr Reid did find much to commend on the site, and I can assure readers that we will do what we can to correct omissions and mistakes as soon as possible, and that we are labouring to find the funding and institutional support to expand the site to cover more case notes and more hospitals. The original vision has changed over the years, but we still hold firm to the commitment to open up access to the recorded information on the lives of some of the most disadvantaged of Britain’s children to all who are interested.