Territorios y conocimientos Aproximaciones, convergencias y divergencias
COMPARACIÓN DIMENSIÓN ESPACIAL DIMENSIÓN COGNITIVA
In the context of growing concerns about medical litigation in the 1970’s, a different methodology for measuring the incidence of preventable patient harm across the hospital population was developed. Because Governments were seeking to explore potential alternatives to negligence-based litigation for patients harmed in healthcare, rather than examining healthcare quality from the clinical care perspective in a specific institution, the overall incidence of harm became more important. The most common methodology used in these studies was retrospective medical record review, but a small number used prospective observation methodologies27, and in one pilot study, used both
24 See eg, Couch NP. Tiney NL. Rayner AA. Moore FD. The high cost of low frequency events –
the anatomy and economics of surgical mishaps. 1981 New England Journal of Medicine, March 12, volume 304(11), pages 634-63. While this prospective study of general surgery only located 36 patients with adverse outcomes due to error in general surgery in a pool of 5,612 patients in a 12 month period, 20 of these died in hospital – 11 directly as a result of the adverse event. There were 56 important errors in these 36 patients, resulting in 61 complications. Two-thirds of the errors involved an error of commission, defined as an unnecessary, defective or inappropriate operative procedure. 5 of the 16 survivors left hospital with serious physical impairments. The average hospital stay was 42 days (1-325 days range) and the total costs for the 36 patients was $US1,732,432. The most frequent complications were intra-abdominal sepsis eg abscess, peritonitis, (15/61), followed by ischaemic non-cardiac injury eg stroke, gangrene (9/61), renal injury (9/61), mechanical lesion of the bowel (8/61), and cardiac injury eg acute myocardial infarction, arrhythmia, congestive heart failure (7/61).
25 See eg, Trunet P. Le Gall JR. Lhoste F. Regnier B. Saillard Y. Carlet J. Rapin M. The role of
iatrogenic disease in admissions to intensive care. 1980 Journal of American Medical
Association, 12 December, volume 244(23), pages 2617-2620. This French study looked at 325 patients admitted to the ICU between 3 August 1978 and 3 August 1979. 41 (12.6%) were admitted from iatrogenic disease. While many had other serious conditions, the admission of 19 patients (46.3% of those with iatrogenic disease) arose from iatrogenic disease caused by avoidable therapeutic or technical errors. The disease caused death in 8 case and was life- threatening in 13.
26 See eg, Lakshmanan MC. Hershey CO. Breslau D. Hospital admissions caused by iatrogenic
disease. 1986 Archives of internal medicine, October, volume 146, pages 1931-1934. This US study studies 834 admissions, which showed 47 distinct iatrogenic events resulting in 45 admissions (5.4% of all admissions). 35 cases were caused by medications, 9 by procedures, 1 each by radiotherapy, transfusional therapy and hospital acquired infection. Around 50% of these admissions were considered avoidable.
27 See Table 1.2 later in this Chapter, National French Adverse Events Study (ENEIS) (Study table
note 9), French Pilot Comparative Methods Study (Study table note 10) and Iberoamerican Study of Adverse Events (IBEAS) (Study note table 14) – these studies used medical records as part of their data source but as part of the observation process. They used similar screening tools as the Medical Record Review studies.
Chapter 1 – Preventable Patient Harm
methodologies on the same patient population.28 The commonalities across both kinds of studies were the use of a similar definition for an “adverse event” and the choice of a sample, the results from which could be generalised across the country or state’s
hospital population.
The first three retrospective medical record review studies29 took place in such a policy environment. There was a perceived litigation “crisis” and doctors (in particular) were calling for action to reduce their medical indemnity premiums. Because of their broader brief, these studies each involved doctors, lawyers, indemnity funders, governments and, in the case of the Professional Indemnity Review, healthcare consumers, in the policy development processes surrounding these studies.
While these three studies were established as responses to a perceived “litigation” crisis, the evidence in each case showed a different picture. In both the US and Australia, the premium rises, which were seen as prima facie evidence of a “litigation crisis”, were not caused by a steep rise in the incidence of litigation. Rather, they were caused by the need for a financial “catch-up” following consistent, long term underfunding of this form of cover through inappropriate price setting by indemnity insurers and other funders, as well as poor investment outlooks domestically or internationally. 30
What became clear from the negligence-based studies was that regardless of the level of litigation, the frequency of harm to patients was much greater than the frequency of litigation against doctors. For example, the Harvard Study showed in its Litigation Claims Study that there were 7.6 adverse events to every incident of litigation in New York in the study year31. Similarly, in the Californian study, about 17.0% of the
28 See Table 1.2 later in this Chapter, French Pilot Comparative Methods Study (Study table note
10).
29 These were the Californian Medical Insurance Feasibility Study, the Harvard Medical Practice
Study and the Quality in Australian Health Care Study conducted as part of the Review of Professional Indemnity Arrangements for Healthcare Professionals (the Professional Indemnity Review), discussed below in Table 1.2.
30 See, for the United States: Danzon PM. Medical Malpractice: theory, evidence and public policy
1985 Harvard University Press, Cambridge (Massachusetts): Part II, page 85 and following; and for Australia: Review of Professional Indemnity Arrangements for Healthcare Professionals. Compensation and Professional Indemnity in Healthcare – Final Report. November 1995 Australian Government Publishing Service, Canberra (PIR Final Report): paragraphs 9.142 and following; and Baker T. The medical malpractice myth. 2005 University of Chicago Press, Chicago, especially chapters 1-4.
31 Localio RA. Lawthers AG. Brennan TA. Laird NM. Hebert LE. Petersen LM. Newhouse JP.
Chapter 1 – Preventable Patient Harm
potentially compensable events (PCEs) identified were assessed to show evidence of negligence32. Further, in the Harvard Medical Practice Study, the study authors argued there was not a good correlation between litigated cases and harm where negligent care had occurred. For example, of the adverse events where negligence was determined to exist in the study, only 2% of these cases had given rise to a claim.33 Some
commentators have argued that the Harvard Study’s decisions on negligence were very conservative and based only on the medical record content, rather than the evidence that might be available in a negligence action. Therefore, they argue that this figure grossly underestimates the correlation between negligence and litigation34. However, like the Harvard Study, the Professional Indemnity Review found only a small number of negligence claims were being made,35 especially compared to the incidence of patient harm through adverse events, as disclosed in the Quality in Australian Health Care Study.
While there have been slight variations in the definition of an adverse event in the various modern adverse event studies, they all share common features36: the occurrence of an unintended injury, caused by healthcare management rather than the underlying disease process, the consequences of which were sufficiently serious to result in
– results of the Harvard Medical Practice Study III. 1991 New England Journal of Medicine, volume 325(4), pages 245-251 (HMPS article III): at page 248. This differs slightly from the estimate in the HMPS Report, due to additional cases identified in the follow-up (an additional 580 records, of which 12 had filed claims against 18 providers). For comparison, see Harvard Medical Practice Study. (HMPS) Patients, doctors, and lawyers: medical injury, malpractice litigation, and patient compensation in New York: the report of the Harvard Medical Practice Study to the State of New York. 1990 Harvard Medical Practice Study Boston. [HMPS Report] A full copy of the scanned Report is available electronically through the New York State Library, via WebCat Call no: HEA 302-4 PATDL – 90 33997, as a 4 part download. Accessed 27 September 2006.page 7-28, where a ratio of 9:1 (or around 11%) was found.
32 Mills DH. Report on the Medical Insurance Feasibility Study. Sponsored jointly by the
California Medical Association and California Hospital Association. 1977 Sutter Publications, San Francisco [referred to as “the Californian Study”]: Table 66, page 98.
33 Chapter 7 of the HMPS Report at note 31- sets out the results of this work, but more distilled
data can be found in the peer reviewed report of this work in the HMPS article III, also at note 31.
34 Baker T. Reconsidering the Harvard Medical Practice Study conclusions about the validity of
medical malpractice claims. 2005 Journal of Law, Medicine and Ethic,. Fall, volume 33(3), pages 501-514.
35 See PIR Final Report at note 30: paragraphs 2.54-2.103 and 7.8-7.12.
36 The definition and two stage medical record review methodology came first from the Californian
Study’s definition of a potentially compensable event (PCE). Mills DH. Report on the Medical Insurance Feasibility Study. Sponsored jointly by California Medical Association and California Hospital Association. 1977 Sutter Publications, San Francisco [referred to as “the Californian study”]: page 8 (PCE) and pages 23-33 for the methodology.
Chapter 1 – Preventable Patient Harm
prolongation of the patient’s otherwise expected hospital stay, temporary or permanent disability or death. All studies required doctors to determine:
• causation across a scale: only those events were counted where causation was more likely than not or greater to have been caused by medical or healthcare management; • preventability across a scale – those were considered to be preventable where it was
more likely than not to have been preventable with current clinical knowledge at the time of the occurrence; and
• disability outcomes for affected patients across a scale – if a preventable adverse event required treatment but did not prolong the person’s hospital stay, then it was not counted as an adverse event in most studies.
The earlier US studies, driven by the litigation concerns discussed above, also asked doctors to make judgments on whether the occurrence involved negligence.
The methodology of all the retrospective medical record review studies is similar. A sample of records is selected, generally representative of the system’s patient
population. In some studies, a specific population is oversampled to obtain more data on areas of specific concern, for example, older patients, younger patients, patients in a specific specialty. Where there is oversampling, this is then corrected in the results. The first stage review of the sampled medical records is usually conducted by a nurse or medical records administrator using a screening tool. The screening tool lists events likely to be evident on a medical record, which may indicate an adverse event has occurred37. Examples of these are unplanned readmission to theatre or unplanned transfer to intensive care, unexpected death, adverse drug reactions, and injury in hospital38. The second stage review is done by either one or two doctors, using a very detailed reporting pro-forma (written or computerised in the more recent studies). The pro-forma information has been similar, with some studies requiring a narrative
summary and some not. The grading scales for degree of causation, preventability and disability have been similar across the different retrospective case note studies.
37 The number of criteria have varied between studies – for example, the California Study had 20
criteria, the Australian Study, the Canadian Study and the UK pilot study had 18 criteria, the Harvard Study had 15 and the recent Latin American IBEAS study had 19 triggers.
38 The exact criteria have varied a little between studies, but almost all of the criteria have been
shared. For the list used in the Quality in Australian Health Care Study, see Wilson RMcL. Runciman WB. Gibberd RW. Harrison BT. Newby L. Hamilton JD. The Quality in Australian Health Care Study. 1995 Medical Journal of Australia, 6 November, volume 163, pages 458-471 (QAHCS article): Table 1, page 460.
Chapter 1 – Preventable Patient Harm
Studies using a similar methodology have been conducted across many countries now, and are generally used as a “flag” indicator of the frequency of preventable patient harm across a healthcare system. Table 1.2 summarises the main results of these studies:
Table 1.2: Modern Hospital Population-wide Adverse Event (AE) Studies Study name Place,
Country
Hospital and patient coverage
No in sample Key injury results Disability results Other interesting findings
California Medical Insurance Feasibility Study *1
California, USA
all patients discharged from non-federal, short-term, general hospitals in California in 1974 20,864 hospitalizatio ns 1,155 PCEs altogether, 970 attributable to 1974 or 4.65% admissions 80.0% temporary disability, 6.5% minor permanent. 3.8% major permanent, 9.7% death
75% of deaths occurred in patients who were not expected to die within one year of the admission
Harvard Medical Practice Study *2
New York, USA
Acute care, non-psychiatric hospitals, Sample weighted to reduce % of older people and increase % of specialties thought to be at high risk of AEs then corrected in results
30,121 admissions
1,278 AEs and 306 negligent AEs identified, 1,133 AEs were within defined time limits, giving an AE rate of 3.7%, with 27.6% of AEs involved negligence 56.8% recovered in 1 month, 16.5% 1-12 months, 6.5% permanent disability, 13.6% death, 6.6% unknown degree of disability
People over 64 (27% of hospital population) had higher incidence of AEs (43% of AES in this group) and higher proportion of negligent AEs (52%). 48% of all AEs associated with operations, 52% non-operative.
Quality in Australian Health Care Study *3
Australia 28 hospitals - Stratified sample of general non- psychiatric hospitals & hospital population
14,179 admissions
16.6% admissions associated with an AE, 8.3% highly preventable, Annual AE rate 13% 46% recovered within 30 days, 30.5% between 1-12 month, 13.6% permanent disability, 4.9% resulted in death.
Agreement between 2 doctors at Stage 2 occurred in 67.9% of records, 32.1% required a third review - 69% of these were disputes about whether there was an AE disclosed. Utah-Colorado Medical Practice Study *4 Utah & Colorado, USA 15 hospitals in Colorado, 13 in Utah, excluded veterans, psychiatric, rehabilitation and drug and alcohol hospitals
14,700 records
169 AEs in Utah, 418 in Colorado, Overall AE rate 2.9%, negligence rate in Utah 27.5% AEs, 32.6% in Colorado 53.3% minor temporary, major temporary 31.6%, permanent disability 7.5%, grave disability or death 7.5%
Used only one reviewer at Stage 2 - use additional quality checking to address reliability issues. . Elder Care study component showed AEs twice as likely 65 and over, and death rate for preventable AEs over 65 was over 10%.
British Pilot Study *5 London, England
2 hospitals - index admission in 2 months of 1998 1,014 random files 110 patients suffered 119 AEs, 11.7% AE rate, 48% preventable 66% minimal impairment, 19% moderate impairment, 6% permanent impairment, 6% death.
Patients with AEs on average 68.5 v 47.5 years without. 119 AEs resulted in 999 extra bed days. Quality of medical records was inadequate in about half.
New Zealand Adverse Event Study *6
New Zealand
13 acute care public hospitals with over 100 beds - no psychiatric or rehabilitation admissions 6,579 admissions 13% prevalence rate, 11.3% AE rate with 10.4% occurring in public hospitals, 37.1% preventable 7.9% recovered in 30 days, 19.0% in 1-12 months, 10.2% permanent impairment, 4.5% death, 4.7% unable to be determined from record.
AEs more common over 65. When the AE occurred outside hospital, there was a greater degree of disability and death.
Study name Place, Country
Hospital and patient coverage
No in sample Key injury results Disability results Other interesting findings
Danish Pilot Study *7
Denmark Random sample from Danish National Patient Register from 17 different acute hospitals
1,097 admissions
176 AEs found in 114 admissions, prevalence rate 9.0% of admissions, 40.4% preventable
Most effects were transient, but in 26.3% of AE admissions, death or permanent disability resulted. Canadian Adverse Event Study *8 British Columbia, Alberta, Ontario, Quebec, Nova Scotia
Inpatients in acute facilities within 250 kms of provincial research centre, with >1,500 inpatients in 2002, 24 hr emergency department - one teaching hospital, one large community hospital .100 beds, 2 small community hospitals < 100 beds 3,745 admissions 255 charts contained 289 AEs. 7.5% AE rate, 36.9% preventable 55.7% recovered in less than 1 month, 12.5% in 1- 12 months, 5.2% suffered permanent impairment, 15.9% death, and in 10.4% disability couldn't be determined. 1.6% of admissions died from an AE
255 AE patients required additional 1,521 days in hospital directly related to the AE, Errors of omission in medicine caused 57.1% of medical AEs, but 50-50% in surgical cases
French Pilot Comparative Methods Study *10 Aquitaine, France Patients in 37 wards in 3 public and 4 private hospitals
778 patients 241 AEs in 174 patients - 70% had 1, 22% had 2 and 8% had 3 AEs. Overall AE rate was 31.0%
Not obtained- purpose of study was to test methods for ENEIS below
This study compared cross sectional (data gathered on one day), prospective (data gathered during hospital stay), retrospective record review method over same patient sample. Only 25% of all AEs were detected by all methods, and less than 4% of preventable AEs. Retrospective and prospective performed best. National French
Adverse Events Study (ENEIS) *9
Aquitaine, France
117 hospitals 8,754 patients 450 AEs - 181 preventable - 6.6% AEs per 1,000 bed days during hospitalisation, 37.2% preventable, in addition 4% of admissions were caused by AEs - 45% were preventable Temporary disability/ prolongation of stay 66.2% of AEs; 29.3% Serious disability (permanent or life threatening); 1.7% death
This study showed higher AE rates with older, more fragile patients. Prospective study method used.
Spanish Adverse Events Study (ENEAS) *11
Spain Patient had to have been an inpatient in one of 24 hospitals in the sample representing all Spanish hospitals by size for at least 24 hours and been
discharged between 4-10 June 2005.
5,624 records 9.3% incidence of AEs. (525)- 473 occurred in hospital. In 105 AE caused the admission 45% were considered of low seriousness (43.5% preventable), 39% of moderate seriousness (42% preventable), 16% serious AEs (41.9% preventable)
Patients over 65 with other risk factors were 2.5 times more likely to suffer an adverse event than a younger person without risk factors.17.7% of patients with an AE had more than one AE.
Study name Place, Country
Hospital and patient coverage
No in sample Key injury results Disability results Other interesting findings
Dutch Adverse Events Study *12
Holland 21 of 101 Dutch hospitals- 4 university, 6 tertiary teaching, 11 general -3,983 deceased patients and 3,943 discharged patients in 2004
7,926 admissions
5.7% of admissions had one or more AE, 40%
preventable
12.8% resulted in death or permanent disability; 10.7% of deceased hospital patients had an AE, and in 4.1% of hospital deaths, a preventable AE
contributed to the death.
Proportion and impact of AEs increased with age. More than 50% were related to surgery.
OIG Medicare Adverse events study *13
USA Statistically valid sample of Medicare patients discharged in October 2008 780 beneficiaries 13.5% suffered AE and another 13.5% had temporary disability requiring treatment but not defined as an AE; 44% of both were considered preventable. Estimated direct annual care costs of preventable harm in 2009 was $4.4B - 3.5% of inpatient expenditure.
0.6% had a National Quality Forum Serious Reportable Event, 1.0 had a Medicare Hospital Acquired Condition (HAC) event, and 13.1% experienced one of the four most serious
categories of patient harm - 1.5% AE contributed to death.
86% of harm not reported in ordinary incident monitoring systems - 62% because staff did not consider it reportable, 25% because while they normally reported these kind of things, this time they didn't. 28% of beneficiaries who had an AE also experienced temporary harm. Projection of death rate to Medicare beneficiary hospital population is 15,000 patients per month.
Iberoamerican study of adverse events (IBEAS) *14 Argentina, Colombia, Costa Rica, Mexico and Peru 58 hospitals across 5 countries - 2000 patients per country during the study week AE had to be present on day of observation
11,379 inpatients
1,191 patients had at least one AE, point prevalence rate 10.5%, 60% of which were considered preventable
28% of AEs caused disability, 6% associated with death of the patient, 64.7% of AEs prolonged hospital stay on average of 16 days
Used medical record review tools but observations only related to one day in hospital so data not comparable with other studies - sought point prevalence instead. 40% of AEs related to hospital acquired infections, and 1.5% of admissions were caused by AEs Eastern Mediterranean and African Adverse Event Study *15 Egypt, Jordan, Kenya, Morocco, South Africa,