# What Paper Charts Did Terribly

*Four ways the folder failed, and what happened to each when the paper went*

Kanwar Partap Singh Gill, MD · November 7, 2026 · Becoming a Doctor

In 1994 three researchers sat in on one of the oldest rituals of a teaching hospital: residents in internal medicine presenting their outpatients, one after another, to the attending physician who would hear the story and approve the plan. They recorded 168 consecutive presentations, transcribed them, and listened for a particular kind of sentence, the question that showed someone looking for something in the chart and not finding it. Such questions turned up in 81 percent of the cases. The paper that reported this, by Paul Tang and his colleagues, is quiet in tone, and its conclusion is a sentence of almost bureaucratic restraint: the record system they studied "did not provide effective access to patient information" for physicians making decisions in the clinic. Translated out of the language of systems, it says that four times in five the doctor in the room wanted something the folder held, or ought to have held, and could not get it.

I begin there because this essay is a charge sheet, and charge sheets should begin with evidence. It has a companion on this site about [what paper charts did better](../what-paper-charts-did-better/index.html), and the two are meant to be read together, but this one is written without affection. A folder is easy to love at a distance. The distance flatters it in the way it flatters most retired instruments, by remembering the times it worked and forgetting the afternoons spent looking for it. In the flagship essay on this site, [The Primary Care Office After the EHR](../the-primary-care-office-after-the-ehr/index.html), I listed paper's failings in six short lines and then called them honest, because they were visible. A separate essay, [The Honest Gap](../the-honest-gap/index.html), puts that claim under pressure. What I want to do here is what a list cannot: take the failures apart one mechanism at a time, because they were never one failure and they did not have one cure.

The records among which I trained in Amritsar, until 2002, were overwhelmingly paper; the flagship essay says as much, and beyond that public fact this essay claims no memory of its own. It does not reconstruct a ward, a record room or a missing folder. The mechanisms it describes were general, the evidence for them was mostly gathered in other countries, and the cases that show them at work are imagined and are introduced as such.

There were at least four mechanisms. A paper record is a single object, so it exists in one place at a time. It is written by hand, so its meaning depends on whether a stranger can decipher someone else's haste. It belongs to an institution, so a patient treated in three hospitals has three records that know nothing of one another. And it moves by being carried, so a result that arrives in the building must still be physically brought to a mind that will read it. A fifth is less a failure than a limit: paper protects privacy only as well as the doors, drawers and dustbins around it. Each mechanism did harm in its own way and asked for its own remedy. The interesting question, a quarter of a century into the electronic record, is which remedies arrived, which arrived in part, and which failures simply changed their clothes.

## One Object, Many Claimants

Scarcity is the most physical of the failures and the easiest to forget, because it was so ordinary that nobody thought of it as a failure at all. A paper chart is a thing. It has a weight and a location, and at any given moment it is in exactly one location. The outpatient clinic wants it for this morning's visit. The ward wants it because the same patient was admitted overnight. Radiology wants it so that the films can be compared with the last report. The records department wants it back on its shelf. A consultant across the hospital would like to read the history before going to see the patient. A clerk in another office needs it to answer a letter. Every one of these claims is legitimate, and they cannot all be satisfied at once. The folder ends up wherever the most recent or most insistent claimant left it, and the next person to need it begins with a search.

Institutions built a whole craft around this scarcity, and the people who practised it deserve more credit than nostalgia gives them. Records staff kept registers of where each folder had gone and left tracer cards in the gaps on the shelves; porters carried folders between floors; ward clerks knew which pile on which trolley held what. These systems often worked, and their working was invisible, which is the fate of most good clerical labour. When they failed, they failed in a characteristic way, and the characteristic workaround made things subtly worse. Clinics that could not count on the main record kept their own: a thinner folder in a drawer, with copies of letters and the clinic's own notes. The duplicate solved the problem of availability by creating a problem of agreement. Now there were two records, each incomplete in a different way, and nothing in either told the reader what the other contained. A second copy of a scarce thing is not abundance. It is a second scarcity.

How often did this matter? The best estimates come from the years just before electronic records spread, and they come from clinicians' own reports, which is a limitation worth keeping in view. In 2003, Peter Smith and colleagues asked 253 clinicians in 32 primary care clinics in Colorado to complete a short questionnaire after every visit in a half-day session, saying whether important clinical information had been missing. Across 1,614 visits, they said it had in 13.6 percent: laboratory results in about six in a hundred visits, letters and dictated reports in about five, radiology results in about four. More than half the time, the clinicians believed the missing information was somewhere outside their own clinical system. In 44 percent of the visits with something missing, they judged its absence at least somewhat likely to harm the patient. These are perceptions, not audited outcomes, and the authors said so. But one pattern in their data deserves more attention than it usually gets. Missing information was reported more often for new patients and for patients with many medical problems, and the odds rose with the number of problems.

That is the hard-headed point about scarcity. It did not fall evenly. It fell hardest on the patients whose care was most spread out and whose histories most needed to arrive with them: the newcomer, the person with five diagnoses and four doctors, the patient whose last admission was across town. A system whose failures concentrate on the sickest and the least settled is worse than inefficient: it is regressive, in the strict sense that it taxes most heavily those least able to pay. Another essay here sorts out the different things a doctor can mean by [I can't find it](../where-does-this-fact-live/index.html). For this essay the cause matters more than the taxonomy, and the cause was physical. The information existed. It was simply somewhere else.

## The Reader Pays

Whether doctors really write worse than other people, and what a recognised hand does and does not prove, are questions taken up in [The Handwritten Medical Chart](../the-handwritten-medical-chart/index.html). The question here is narrower: what illegibility did to the person on the other end. Handwriting is a transfer of cost. The writer saves seconds by writing fast, abbreviating freely and letting the last letters of a word trail into a line. The reader pays for those seconds later, in minutes of squinting or in a misreading, and the payment falls due at the worst possible moment: when the writer has gone off duty and the reader has less context than the writer had. The bargain is struck by one person and honoured by another, who was never asked.

In November 2002 a group of physicians at a general hospital in southern Spain published a small study with a plain title, "Illegible handwriting in medical records." Two observers independently scored the legibility of the clinical histories in a sample of 117 case notes, with a third to settle disagreements. In 18 of them, 15 percent, the defects were such that the whole history was unclear; they were commonest in records from surgical departments. The authors drew a conclusion that was blunt for a medical journal: if clinicians could not be persuaded to write legibly, "the solution must be an accelerated switch to computer-based systems." The study was small and local, and nobody should turn it into a rate for every hospital of its time. Its value is that it measured something everyone complained about and almost nobody counted.

Abbreviation is the same transfer in a more respectable form. A shortened word works perfectly among people who share its key, and the trouble is that a hospital is many circles of people with different keys. The same letters can mean one thing in one department and another down the corridor. Some abbreviations collide with numbers: a *U* for units can be read as a zero, and a trailing zero after a decimal point can turn one milligram into ten. These became the standard examples on the lists of forbidden abbreviations that safety bodies later published, and they were chosen for a reason. A misread word in a history is a nuisance; a misread quantity in an order is a dose.

It is often said in paper's defence, and I have said it myself, that illegibility was at least honest: you knew when you could not read a word. That is true only of the illegibility that announces itself. The more dangerous kind does not. A word that is half legible is read as the nearest word the reader expected, and it arrives in the reader's mind as clearly as if it had been printed. The misreading carries no mark of doubt. Nobody writes *I guessed* beside a guess. In the act of reading, the visible gap of a scrawl could become a confident error, which is exactly the failure that paper's admirers reserve for screens. The honesty of an illegible page depended entirely on the humility of its reader.

Typing ended this failure almost completely, and it remains the clearest single gain of the whole transition. A typed dose cannot be mistaken for a different dose because of the writer's hand. But typing also showed that legibility was never the whole of reading. A typed note can be perfectly legible and practically unread, long enough that the one sentence that matters is lost among sentences that do not, a problem another essay takes up under the heading of [the note that grew longer and said less](../the-note-that-grew-longer-and-said-less/index.html). Handwriting charged the reader in deciphering. Templates charge the reader in sifting. The reader still pays.

## The Page in the Wrong Folder

One failure belongs to paper so completely that it is rarely listed among the others, and it is worth a short section of its own because its successor is still with us. A paper record carried its patient's identity on its cover. The sheets inside carried it only as well as someone had written or stuck a name and number on each of them. A continuation sheet, a laboratory slip, a letter from another hospital, an electrocardiogram tracing folded into quarters: each was a loose object until it was filed, and once filed it belonged to whichever folder it had been put in. The folder vouched for its contents, and readers trusted the folder.

A misfiled page is a double failure, and both halves are silent. In the right folder it leaves no hole, because a folder does not know what it ought to contain; the result simply is not there, and its absence looks like nothing at all. In the wrong folder it is worse than absent. It is present, under another person's name, where a clinician in a hurry may read a stranger's result as his patient's, and where a stranger's private history is now open to anyone who opens that folder. Only a reader who checks the name on every sheet, every time, catches the intruder, and nobody reads every sheet of a thick chart every time.

Here too the failure changed its clothes when the record went electronic, and here the change can be measured. In a study published in 2013, Jason Adelman and colleagues looked for a tell-tale pattern in computerised order entry at four hospitals: an order placed for one patient, promptly retracted, and then placed by the same clinician for a different patient. When the researchers checked a sample of such events, about three in four turned out to be genuine wrong-patient orders, and on that basis the authors estimated that 5,246 electronic orders had been placed on the wrong patient in those hospitals in 2009. They then tested two remedies in a randomised trial covering more than 900,000 ordering sessions. An alert asking clinicians to verify the patient's identity reduced the odds of the pattern modestly. A function requiring them to re-enter the patient's identifiers before ordering reduced them by about two-fifths. The page in the wrong folder had become the order on the wrong screen. What survived the change of medium was the underlying mechanism: identity attached to a container rather than to each act, and a busy person's trust that the container is the right one. What the change made possible was counting the failure and testing a cure, which paper never allowed.

## A Result Looking for a Reader

The mechanism of carrying deserves the closest attention, because it outlived the paper that created it.

Consider a hypothetical example. A blood test is drawn in an outpatient clinic on a Thursday morning. The specimen reaches the laboratory the same day; the result is printed on Friday afternoon, on a slip of paper bearing the patient's name and number. The slip goes into the internal post. On Monday it reaches the clinic, where a clerk, following the clinic's rules exactly, files it in the patient's folder. The folder goes back to the records shelf. The patient's next appointment is in three months. The result is abnormal in a way that would have changed something had anyone seen it that week.

Follow the chain again and notice what is missing. Every step was carried out correctly. The specimen was not lost, the slip was not misaddressed, the clerk did not neglect a duty. At no point did anyone fail to do his job, and at no point was anyone's job to read the slip and decide. *Filed* is a verb of storage. The system had a place for every piece of paper and no place for a decision. The result was delivered with complete success to a folder, and a folder does not think.

Paper made this failure easy because, on paper, the evidence of delivery and the evidence of reading looked identical. A result slip in a folder looks the same whether a clinician pored over it or nobody has glanced at it. The best paper systems knew this and added a small ceremony: the clinician initialled and dated the corner of each slip before it could be filed. The initials were paper's receipt of acknowledgment, and they were a real improvement. Their limitation was that they recorded the passage of a pair of eyes. They did not record what the eyes concluded, whether anything was ordered, or whether the patient was told.

Herman Melville ended "Bartleby, the Scrivener" (1853) with a rumour that his narrator cannot shake off: that before coming to the law office, the silent copyist had been a clerk in the Dead Letter Office in Washington. "Dead letters!" the narrator exclaims, "does it not sound like dead men?" He imagines the clerk taking from the folded paper a ring meant for a finger that now moulders in the grave, a bank-note sent in charity to someone who no longer hungers, pardon for those who died despairing. And he gives the image its final turn in a sentence I cannot improve on: "On errands of life, these letters speed to death." What is precise in Melville's picture is that the letters are not lost. They are in a building, handled, sorted, eventually burned by the cart-load. Their failure is that no living reader received them in time. A result without a reader is a dead letter in exactly that sense, and the filing cabinet of a busy clinic could be, on a bad week, a small dead letter office with a pleasant receptionist.

The evidence that this was common, and that physicians knew it, is better than the evidence for most of paper's failures, because test results can be counted. In a survey published in 2004, Eric Poon and colleagues asked 262 physicians in fifteen internal medicine practices affiliated with two large teaching hospitals in Boston about their handling of results. Eighty-three percent reported at least one delay in reviewing a result in the previous two months. They spent, on their own estimate, an average of 74 minutes a clinical day managing results, and only 41 percent were satisfied with how they did it. Asked what they wanted, they named tools to help them write to patients about results, to prioritise their work, and to track test orders to completion. That last phrase is the whole problem in five words. What the physicians wanted was not faster delivery. They wanted to know when a thing they had started was finished.

Another essay on this site follows the result that is still pending when a patient leaves hospital, and the question of who owns it once she has gone; I will not repeat it, except to say that [The Handoff](../the-handoff/index.html) and this section describe the same failure on either side of the hospital wall. The result I have imagined here was born and died inside one building, in the care of people who were all doing their jobs.

## Fragments Under Different Numbers

In 1863, in the third edition of her *Notes on Hospitals*, Florence Nightingale set out a plan for uniform hospital statistics and explained why she needed it. She wanted to know whether hospitals as then conducted were saving lives or wasting them, and she could not find out. "In attempting to arrive at the truth," she wrote, "I have applied everywhere for information, but in scarcely an instance have I been able to obtain hospital records fit for any purposes of comparison." Her remedy, appended to the book, was a set of model tables and a uniform nomenclature, a common list of names for diseases and operations to be used by every hospital in filling up its returns. She had understood that records kept separately, under different names and in different forms, do not add up to knowledge, however carefully each is kept.

Nightingale was asking about institutions, but the same fragmentation runs through the history of every patient who has been treated in more than one place. Each hospital numbered its patients in its own way. Each filed under its own conventions. Each wrote the history afresh, sometimes from the patient's account and sometimes from a letter that may or may not have come. A Punjabi name, transliterated into English by a different clerk in each building, could be spelled three ways in three folders, with the family name sometimes written where the given name belonged; a search by name in one building would not find the patient's record in another even if anyone had thought to search there. No index joined these records. The only thing they had in common was the person, and the person was rarely asked to serve as the index, though she was the only one who could have.

There is an illusion here that deserves a name. A thick chart persuades. Its weight suggests completeness, and a clinician holding a folder two inches deep feels, reasonably enough, that the history is in his hands. But the thickness of a folder measures one institution's attention, not the patient's life. The two-inch chart may hold forty visits to this hospital's clinic and nothing about the operation in another city, the diagnosis made by a private specialist, or the drug started by a doctor the patient saw on holiday. The most important page was often the one in another building, and nothing in the thick folder admitted its absence.

In India, the professional-conduct regulations that the Medical Council of India notified in March 2002 addressed records directly. In the text as the national regulator now publishes it, a physician must keep the records of his indoor patients for three years from the start of treatment, and if a patient, an authorised attendant or a legal authority asks for them, the request is to be acknowledged and the documents issued within 72 hours. Read as a remedy for fragmentation, the rule is revealing. It does not join hospital to hospital. It gives the patient the right to a copy, and so makes the patient the means by which records cross from one institution to the next. The same section ends with a sentence that reads, at this distance, like a small prophecy: "Efforts shall be made to computerize medical records for quick retrieval." The regulator had identified the problem as retrieval and the hope as the computer. Both identifications were correct. Neither, as it turned out, was sufficient. How a document actually crossed between institutions in those years, by post, courier, telephone, fax and the patient's own hands, is the subject of a companion essay, [The Fax Machine as Medical Device](../the-fax-machine-as-medical-device/index.html).

## What Paper Exposed

Privacy is the failure that paper's defenders most often claim as a strength, and in one sense they are right. A folder on a shelf cannot be read from another continent, copied by the thousand in an afternoon, or sold in bulk. Paper's privacy was the privacy of scarcity and obscurity. But scarcity and obscurity are not the same as protection, and paper's exposures, though smaller in scale, were constant and largely unrecorded.

They were physical, and they followed the object. The case notes kept at the bedside, in what many hospitals in the subcontinent call the bed-head ticket, could be read by anyone at the foot of the bed in an open ward. The folder carried along a public corridor could be opened by whoever carried it. The photocopier made duplicates that nobody tracked. The fax tray, as the companion essay describes, received pages that anyone passing could pick up. And at the end of its life, paper became waste.

Punjabi has an exact word for that last condition. ਰੱਦੀ (*raddī*), which Platts's old dictionary of Urdu and Hindi glosses as rejected, used and done with, and as waste paper, names the stacks of old newspapers and exercise books that households sell by weight, and that return to the street as cones and bags in which a vendor hands over roasted gram. I make no claim about any hospital's handling of its old records. The point is structural. Paper's confidentiality ended at the moment it was judged to be waste, and depended entirely on whoever made that judgment and on what they did next. A medium whose privacy is guaranteed only until someone decides it is rubbish is not private by nature. It is private by custody.

The same regulations of 2002 put the duty where it belonged. Under the heading "Patience, Delicacy and Secrecy," they say that "Confidences concerning individual or domestic life entrusted by patients to a physician" should never be revealed "unless their revelation is required by the laws of the State." The duty was the physician's. The paper, however, passed through many hands that the duty did not name, and none of those hands left a trace. That is the deepest limit of paper's privacy, and it is easy to miss: a folder cannot tell you who has read it. A curious porter, a gossiping visitor and the treating physician all leave the same mark on a page, which is none.

Here the electronic record made a real advance. In the United States the federal security rule requires systems holding health information to keep audit controls, records of who opened what and when, and the existence of such a log changes behaviour as well as permitting investigation. A record that knows who has read it is a different moral object from one that does not. The new medium brought its own exposure, of course, which is scale: a breach can now take in more records in a night than a careless records room could expose in a century. Neither medium is private by nature. Privacy is a practice, and each medium makes some practices cheap and others expensive.

## The Failure Changes Its Clothes

It would be convenient to end with paper convicted and the screen acquitted. The evidence does not allow it, and the most interesting piece of that evidence comes from a study of test results.

In a study published in 2009, Lawrence Casalino and colleagues reviewed the records of 5,434 randomly selected patients aged 50 to 69 in 23 primary care practices in the United States, looking for abnormal results and for evidence that the patient had been told. They found apparent failures to inform the patient, or to document having done so, in 7.1 percent of 1,889 abnormal results, with a range across practices from none at all to more than a quarter. Practices with better processes for managing results had lower rates. And then the finding that matters here: practices using what the authors called a partial electronic record, paper progress notes with electronic results or the reverse, had higher failure rates than practices with no electronic record at all, and higher than practices whose electronic record held both notes and results. The study was observational, it relied on what the chart documented, and a failure to document telling a patient is not proof that nobody told her. But the shape of the result is striking. The danger was greatest at the seam.

That is the lesson I would draw from the whole transition, and it applies well beyond test results. Failures live at joints: between paper and screen, between one system and the next, between the person who orders and the person who reads. A hybrid record doubles the number of places a fact can be and halves the certainty about which place is current. The Colorado study points the same way from the other side: clinicians who reported having full electronic records reported missing information less often. The pure forms were each safer than the mixture.

So, mechanism by mechanism, what happened? Scarcity was largely cured. A record that can be opened simultaneously in the clinic, the ward and the radiology department has abolished the search for the folder, and that alone must have spared patients a great deal. Illegibility was cured, at least for handwriting. Privacy gained an audit trail and acquired a new kind of exposure. Fragmentation was cured within a single system and survived, almost untouched, between systems, which is why so much information still crosses institutional walls by the oldest electronic channel in the building. And routing, the failure of the result looking for a reader, was transformed rather than solved. The result now arrives instantly and infallibly, in an electronic inbox where it waits among hundreds of others, as another essay here describes in [The Inbox That Never Sleeps](../the-inbox-that-never-sleeps/index.html). Delivery became perfect. Reading stayed exactly where it was, in the attention of a tired person with too many things to read.

There is a further point that the Casalino study makes quietly. Its practices' own processes for handling results mattered alongside the kind of record they kept. Some of paper's failures were failures of paper. Others were failures of organisation that paper merely made easy, and they can survive in any medium that does not change the organisation. A clinic whose rule was *file the slip* became, in its electronic form, a clinic whose rule was *clear the inbox*, and the two rules share the same flaw: each defines the task as disposal of the document, not completion of the work the document requires.

## A Fair Trial

A charge sheet invites a defence, and paper's defenders would plead anachronism. It is easy, the defence would say, to convict an old instrument of failing to do what a newer one does; one might as well blame a bullock cart for not being a train. The people who kept paper records worked within limits nobody could then remove, and many of them worked brilliantly. A judgment passed in 2026 on the records of 1994 or 2002 is a judgment made with hindsight, by someone who no longer has to find the folder.

Part of this defence should be granted outright. The clerks, porters and ward staff who made paper systems function were not the failure, and much of what I have described happened despite their work, not because of it. Most of the time, most folders were found, most results were read, most pages were in the right place. Even the most striking figure in this essay needs reading with care: Tang's 81 percent counts presentations in which a physician could not find everything wanted, not presentations in which a patient came to harm, and some of what was wanted would not have changed anything. An honest charge sheet admits the limits of its evidence.

But the charge of anachronism fails on its central claim, because the complaints were not invented afterwards. Nightingale made hers in 1863. The Spanish authors who counted illegible histories in 2002 were not looking back on paper; they were working in it, and asking for computers. The Indian regulation of the same year wrote the hope of quick retrieval into the profession's own code. Tang and his colleagues listened to their clinic's questions in 1994 precisely because they could hear, at the time, what the folder was failing to answer. The people who lived with paper knew its failures well. What they lacked was not insight but an alternative.

And the purpose of the charge sheet is to identify mechanisms rather than to convict the past, because mechanisms outlive media. A failure that has been understood only as a property of paper will be declared solved when the paper goes, and then rediscovered, unrecognised, in the new system. That has already happened at least twice in this essay.

## A Return Address for Every Result

What would it mean to design around the work rather than the document? I think it means making the completed clinical task, not the record, the unit by which a system is judged. A result is not finished when it arrives, or when it is filed, or when it is opened. It is finished when a named person has read it, decided what it means for this patient, done or arranged whatever follows, told the patient, and left a trace of all four that the next reader can see. Paper's initials in the corner recorded the first of these. An electronic acknowledgment records the same first step a little more reliably. Neither records the rest unless someone has built the rest.

Melville's letters sped to death because nothing along their way was charged with noticing that they had not arrived. The opposite design is not difficult to describe, though it is laborious to build. Every result, report and recommendation would carry what amounts to a return address: if it is not acknowledged and acted on within a time suited to its urgency, it comes back, to the person who ordered it or to a team that has agreed to watch for silence. The physicians in Poon's survey asked for something like this when they asked to track orders to completion. The success of such a system would show in a number nobody puts on a dashboard: the results that did not die in a folder or an inbox because someone was waiting for them.

Return, finally, to the 168 presentations in 1994 and the questions that revealed a doctor looking for something the chart could not give. Those residents and attending physicians were not careless. They were asking a paper record questions it was not built to answer: where is the last result, what did the other hospital find, what happened since the last visit. A record today answers most of those questions almost instantly, which is a genuine achievement, not to be dismissed by anyone who remembers the alternative. But there was always one question the folder could not answer and the screen, left to itself, answers no better. It is not *where is this?* It is *has anyone read this, and what did they decide?* Paper failed in many ways. That was the failure it handed on intact.

## Evidence and Editorial Note

**Documented findings.** Public biography relied on: MBBS, Sri Guru Ram Das Institute of Medical Sciences and Research, Amritsar, 2002; the author's published account, in the flagship essay on this site, that his training took place among predominantly paper records. Tang PC, Fafchamps D, Shortliffe EH, "Traditional medical records as a source of clinical data in the outpatient setting," *Proc Annu Symp Comput Appl Med Care* 1994:575–579 (observational study at a university hospital clinic; 168 consecutive case presentations by internal medicine residents to attending physicians; in 81 percent physicians could not find all the information they desired; the quoted conclusion): PubMed abstract, PMID 7949993. Smith PC et al., "Missing clinical information during primary care visits," *JAMA* 2005;293:565–571 (253 clinicians, 32 Colorado primary care clinics, 1,614 visits, May–December 2003; missing information reported in 13.6 percent of visits; laboratory results 6.1, letters and dictation 5.4, radiology 3.8 percent of all visits; 52.3 percent located outside the clinicians' system; 44 percent at least somewhat likely to harm; higher odds for new patients and multiple problems; lower with full electronic records; clinicians' perceptions, not validated): PubMed abstract, PMID 15687311. Rodríguez-Vera FJ et al., "Illegible handwriting in medical records," *J R Soc Med* 2002;95:545–546 (117 case notes, a Spanish general hospital; 18, or 15 percent, with legibility defects rendering the whole unclear; more frequent in surgical records; quoted conclusion): PubMed abstract, PMID 12411618. Poon EG et al., *Arch Intern Med* 2004;164:2223–2228 (262 physicians in 15 internal medicine practices affiliated with two Boston teaching hospitals; 64 percent response; 83 percent reporting at least one delay in reviewing results in two months; 74 minutes per clinical day; 41 percent satisfied; desired tools including tracking test orders to completion): PubMed abstract, PMID 15534158. Casalino LP et al., "Frequency of failure to inform patients of clinically significant outpatient test results," *Arch Intern Med* 2009;169:1123–1129 (5,434 patients aged 50–69 in 23 practices; 7.1 percent apparent failures, 135 of 1,889; range 0–26.2 percent; better process scores associated with lower rates; partial electronic records associated with higher rates than no electronic record or a full one): PubMed abstract, PMID 19546413. Adelman JS et al., "Understanding and preventing wrong-patient electronic orders: a randomized controlled trial," *J Am Med Inform Assoc* 2013;20:305–310 (retract-and-reorder measure with positive predictive value 76.2 percent, 170 of 223 events; an estimated 5,246 wrong-patient electronic orders in four hospitals in 2009; trial of 901,776 ordering sessions among 4,028 providers; ID-verify alert odds ratio 0.84, ID-reentry function 0.60): PubMed abstract, PMID 22753810. Florence Nightingale, *Notes on Hospitals*, 3rd ed. (1863), section on hospital statistics, the quoted sentence and the appended nomenclature for filling up tables: Cornell University Library copy, Internet Archive text. Herman Melville, "Bartleby, the Scrivener" (1853), closing paragraphs on the Dead Letter Office, quoted: Project Gutenberg text. Indian Medical Council (Professional Conduct, Etiquette and Ethics) Regulations, 2002, notified 11 March 2002: clauses 1.3.1 (indoor records kept three years), 1.3.2 (requests acknowledged and documents issued within 72 hours), 1.3.4 (quoted sentence on computerisation) and 2.2 (quoted sentence on confidences), from the text published by the National Medical Commission, which incorporates amendments to 2016. ਰੱਦੀ (*raddī*): Platts, *A Dictionary of Urdu, Classical Hindi, and English* (1884), via the Digital Dictionaries of South Asia, with ordinary Punjabi usage. The use of tracer cards in records management, the bed-head ticket as a South Asian term for bedside case notes, the commonly cited misreadings of *U* and of trailing zeros on lists of error-prone abbreviations, and the audit-control requirement of the HIPAA Security Rule: standard reference knowledge, stated in general terms. **Vendor representations.** None. Electronic records and inboxes are described in general terms; no product is named or assessed.

**Recollection and tradition.** The author's training dates and the predominance of paper in it are his published biography; no ward, record room, folder, colleague, patient or incident from his training or practice is described. The Thursday blood test is a hypothetical illustration, introduced as such. Remarks on the transliteration of Punjabi names, on clinic duplicate records and on records staff are general descriptions of how such systems worked, not recollections. The remark on the *raddī* trade describes general usage and makes no claim about any hospital's disposal of records.

**Analysis and forecast.** The division of paper's failures into scarcity, illegibility, routing, fragmentation and the limits of privacy; the reading of scarcity as regressive; the argument that illegibility was honest only when it announced itself; the reading of Melville and Nightingale; the reading of the 2002 regulation as making the patient the carrier of records; the account of the misfiled page as a failure of identity attached to containers; the answer to the charge of anachronism; the view that failures concentrate at seams; and the proposal that results carry a "return address" are the author's interpretation as of November 7, 2026. **Independence.** This essay is not sponsored by any institution, company, health system or technology developer. It is personal reflection, not medical advice; readers seeking care should contact their own physician's office.



## Related reading

- [What Paper Charts Did Better](https://personal.kpsgill.com/essays/what-paper-charts-did-better/)
- [The Honest Gap](https://personal.kpsgill.com/essays/the-honest-gap/)
- [The Smell of an Old Hospital Chart](https://personal.kpsgill.com/essays/the-smell-of-an-old-hospital-chart/)