OUP user menu

What is known about adverse events in older medical hospital inpatients? A systematic review of the literature

Susannah Jane Long, Katrina Fiona Brown, Diane Ames, Charles Vincent
DOI: http://dx.doi.org/10.1093/intqhc/mzt056 542-554 First published online: 7 August 2013


Purpose Large international studies have shown that older hospital inpatients are at particular risk of adverse events. The purpose of this review was to synthesize data from studies designed to assess the scale and nature of this harm, with the ultimate aim of informing the development of new safety and quality measurement tools to facilitate improved hospital care for these vulnerable patients.

Data sources, study selection and data extraction A systematic search using Ovid SP and other sources was carried out, rigorous inclusion and exclusion criteria were applied and quality assessment of included studies was conducted. Data were synthesized to give a picture of the incidence, types, causes, preventability and outcomes of adverse events in older medical inpatients.

Results of data synthesis Nine relevant studies were identified.  A wide range of adverse event incidences were reported, from 5.29 to 6.2% in re-analyses of large adverse event studies, to 60% in studies in which the development of ‘geriatric syndromes’ (e.g. falls, delirium, incontinence) was also considered to be adverse events. Important causative factors other than age included clinical complexity, co-morbidity, illness severity, reduced functional ability and lower quality of care. Adverse events in older people lead to unnecessary interventions with resultant complications and increased length of stay.

Conclusion More work is needed to understand the complex nature of adverse events in older inpatients. We must tailor safety measurement and improvement strategies to address challenges presented by the complexity of the geriatric syndromes and the processes of care encountered by older inpatients.

  • patient safety
  • adverse events
  • older patients
  • elderly
  • geriatric medicine


Patient safety is at the heart of medicine, and in recent decades, following high-profile events and landmark international reports such as ‘An Organisation with a Memory’ in the UK [1] and ‘To Err is Human’ in the USA [2], rigorous academic study has deepened our understanding of the incidence, underlying causes and outcomes of health care-associated harm [3]. Concurrently, the demographics of hospital inpatient populations reflect our increasingly ageing societies. For example, in the UK, the Wanless Social Care report [4] estimated that by 2026 the number of people in England aged over 85 years will have increased by two-thirds, compared with a 10% increase in the overall population. This will inevitably impact on hospital services, as it is currently people over 65 years account for 60% of acute hospital admissions and 70% of bed days in NHS hospitals [5]. The picture is similar across the world, and consequently we need to focus on improving the management of conditions associated with old age, such as frailty and dementia [6], and to improve the safety and quality of hospital care of all patients with these conditions.

Information is, of course, available about the incidence of some of the single types of adverse events affecting older people in hospitals, such as adverse drug events (published rates vary from 31.9% [7] to 37% [8]) or falls (affecting 4.8/1000 bed days in the UK according to incident reporting data [9]), but these give us only a snapshot of single conditions, rather than an overview of hospital-associated harm in this population.

Several large, international, retrospective adverse event studies have been carried out over recent decades with the aim of building a picture of the scale of health care-associated harm in hospital populations of all ages. The majority of these studies used a two-stage retrospective case record review process to identify the occurrence of ‘adverse events’ (discrete, measurable units of harm) [10]. Where the effect of age was investigated, data from these studies demonstrate incontrovertibly that older people experience more adverse events than their younger counterparts in hospital [1020]. In studies where the underlying causes of the higher incidence of adverse events in older people were explored, analysis showed that factors other than age alone, such as co-morbidity and complexity, were important [14, 17, 19].

Although these are useful insights, due to their aims and nature, the large-scale adverse event studies were not designed to specifically look at safety in older medical inpatients. The methodologies used did not address the challenges of defining and identifying adverse events in such a frail patient group, characterized by multiple pre-existing conditions and complex hospital admissions. This is illustrated by the fact that many studies excluded patients who were severely unwell or deemed to be dying. It can be argued, therefore, that they provide limited information which we can use to direct targeted patient safety efforts for older people [21].

The objective of this review was to systematically investigate the published literature for studies that were designed specifically to examine the scale, types, causes, preventability and outcomes of adverse events in older medical inpatients. The purpose of this was to build a more detailed picture of health care-associated harm in this frail and vulnerable group in order to inform future safety measurement and improvement work tailored to these patients.


Search strategy and inclusion criteria

MedLine (including in-process and other non-indexed citations), PsychINFO and EMBASE were searched online through the Ovid SP interface. Handsearching was incorporated by scrutiny of reference lists of articles and contents lists of appropriate journals, and unpublished and ‘grey’ literature sources were examined.

Keywords and MeSH terms were generated around three facets dictated by the review question: adverse events, older people and hospital inpatients, with a wide range of synonyms generated by all members of the review team. The search was limited to abstracts and titles of papers published in English between 1980 and September 2011 (to allow adequate sampling of work carried out since the emergence of patient safety research), with abstracts available. At each stage, sensitivity checking was carried out by cross-checking with five reference papers. The final search strategy was run on 1 September 2011, a summary of which is shown in Supplementary material, Appendix 1.

All articles retrieved by the search were assessed according to six inclusion criteria. Included studies were required to

  1. investigate the occurrence of adverse events according to the accepted definition [10];

  2. provide empirical data regarding the incidence of adverse events in older patients;

  3. examine adverse events occurring during the inpatient stay (not before or afterwards);

  4. be relevant to older people cared for in general or geriatric medical wards;

  5. give an overview of adverse events in general (not single types of adverse events such as falls, adverse drug events or diagnostic errors, because we sought information to aid the development of general safety and quality measurement tools);

  6. describe adverse events relevant to UK practice.

One reviewer applied the six criteria sequentially to all 294 abstracts identified by the search so that any paper that did not meet any of the criteria was excluded. A randomly selected subset of 100 abstracts were also checked by a blinded second reviewer; there was an acceptable level of agreement between reviewers, with a kappa score of 0.71 (P < 0.001). Discrepancies were resolved through discussion, resulting in the progression to the next stage of all abstracts with discordant decisions. At this stage, 89 full papers for all possibly relevant abstracts were retrieved and all inclusion criteria re-applied to these papers. Figure 1 shows a summary of the selection process, which led to nine papers being included for final review.

Figure 1

Summary of study selection process.

Quality assessment, data extraction and data synthesis

The methodological quality of included studies was assessed using a revised version of the MINORS scale [22], a previously validated checklist designed for use in non-randomized studies, shown in Supplementary material, Appendix 2. The scale was modified so that items within it were focused on quality assessment of the studies in terms of their ability to provide information about adverse events in older people.

Data extraction was carried out using two data extraction forms: one to extract methodological characteristics (study design, sampling, definitions and methods of detection of adverse events, etc.) of the studies, and one to extract data from their results. These data were then synthesized and summarized into the following categories:

  • incidence of adverse events and the relationship between age and adverse events;

  • types of adverse events experienced by older people;

  • preventability of adverse events in older people;

  • factors other than age affecting the incidence of adverse events;

  • outcomes of adverse events.


General features and quality assessment

Details of the methodological characteristics of the nine studies that were included in the review are summarized in Table 1. Seven studies were carried out in the USA, one in the UK and one in Canada. There appears to be a relative dearth of recent work in this area—four studies were carried out in the 1980s, two in the 1990s and only three since the year 2000. Two are re-analyses of data from large adverse event studies of all ages described earlier [23, 24]. Sample sizes varied according to the methods used—smaller studies (such as Foreman et al. [31]) tended to involve prospective clinical assessment of patients, with larger studies employing retrospective case record review (as in Thomas and Brennan [24] and Sari et al. [23]) or analysis of central administrative electronic databases [25]. All studies used arbitrary definitions of ‘elderly’, with only one study stating that the age limit used was based on pilot work for those at most risk [26].

View this table:
Table 1

Summary of methodological characteristics of the studies included in the systematic review

When subjected to the modified MINORS score described above, the studies exhibited a wide variation in study quality ranging from 0.52 to 0.81 (maximum score 1.00), with a mean score of 0.69. Full scoring is provided in Supplementary material, Appendix 2.

Definitions and methodologies used

The studies each had different aims and, therefore, provide data on a wide range of aspects of adverse events in older people. The principal aim of six studies was to examine adverse events in older people—in one study to assess the effect of an intervention [27], in another [28] to look at the effect of patient factors on outcomes of adverse events, rather than their effect on adverse events alone, and in another to assess the impact of a suspected underlying cause [25]. Four of the studies [23, 24, 26, 29] allowed comparison between older and younger patient groups, while the remainder solely examined adverse events in older inpatient groups.

Definitions of health care-associated harm vary hugely between the studies, clearly influencing the type of data generated. Some studies [23, 24, 27, 29, 30] employed definitions similar to those used widely in the large international adverse event studies of all ages, while others sought to identify adverse events specifically occurring in older people, such as those encompassed by the phrase ‘symptoms of depressed psychophysiologic functioning’ in Gillick et al. [26] (confusion, not eating, falls, incontinence). Foreman et al. [31] also looked only for geriatric syndromes: acute confusion, pressure ulcers, nutritional problems, adverse drug reactions, falls, transient urinary incontinence. Ackroyd-Stolarz et al. [25] also acknowledged the need to focus on issues particularly relevant to older people and created ‘geriatric-specific screening criteria for adverse events’, as well as looking for more generic adverse events such as procedure-related complications.

All studies used similar subject recruitment and selection processes, usually using consecutive admissions to the hospitals in question. Some investigators attempted to sample the more dependent patients (such as Lefevre et al. [30], who excluded elective surgical patients and those with a short length of stay), while others employed exclusion criteria which may have meant that the frailest patients were not studied (such as Ackroyd-Stolarz et al. [25], who excluded patients with previous hospitalizations in the last 6 months in order to gain a sample more representative of community-dwelling older people).

All studies used some form of case record review or review of administrative data to aid the detection of adverse events, which brings with it its own problems such as hindsight and observation bias. Some supplemented this with patient or staff interviews, or other data collected on a hospital or state level. It is interesting to note that Becker [27] was the only investigator to use official adverse event reporting systems to detect adverse events; perhaps this reflects a perception of the reliability of existing reporting systems in identifying adverse events.

Table 2 shows a summary of the findings of the nine studies identified by this review in terms of the incidence, types, causes, preventability and outcomes of adverse events in older people.

View this table:
Table 2

Summary of study findings

Incidence of adverse events

A wide variation in the incidence of adverse events in older people was reported in these studies. The two studies that had been part of large international adverse events studies, in which data were re-analysed to look at the differences between the old and the young, showed the lowest adverse event rates: 5.29% [23] and 6.2% [24]. In contrast, studies that had been carried out specifically to look at adverse events in older people found a much higher incidence such as in the study conducted by Lefevre et al. [30], in which there were more adverse events than patients overall, with 1.8 adverse events per patient, and Foreman et al. [31], who reported an incidence of 60%. This huge variation reflects the different methodologies employed, particularly in terms of the different definitions of adverse events used and the fact that investigators who were particularly focusing on older people tended to seek out information about conditions such as delirium or incontinence not included in the large adverse event studies which were not designed with older people in mind.

Two studies confirmed the relationship between age and adverse event incidence reported by the large adverse international studies. Jahnigen et al. [29] found that age seemed to be related to the incidence of adverse events independently of disease severity or length of stay and Sari et al. [23] found that each year of life increased the risk of an adverse event by 28%, after adjustment for length of stay.

Types and preventability of adverse events

A wide range of types of adverse events were reported in these studies, which again varied according to the study's definition of adverse events. These ranged from geriatric syndromes such as falls, delirium, pressure sores and incontinence [26, 28, 31] to more ‘conventional’ adverse events such as adverse drug events, hospital-acquired infections and procedure-related complications in the other studies. A summary of the wide range of adverse events that affect older people is given in Table 3, together with the likelihood that they would have been detected using traditional retrospective case record review approaches.

View this table:
Table 3

A summary of the wide range of adverse events experienced by older people in hospital, and the likelihood of their detection in traditional retrospective case record review studies

Lefevre et al. [30] found that 52.1% of the iatrogenic complications experienced by older people were deemed to be preventable, and Thomas and Brennan [24] found a significantly increased incidence of preventable adverse events in older people than younger, whereas Sari et al. [23] did not, after adjustment for other confounders.

Factors other than age affecting the incidence of adverse events

Becker [27] and Lefevre et al. [30] showed that loss of functional ability was independently associated with adverse events, while Thomas and Brennan [24] reported that clinical complexity probably accounted for the observed association with age, rather than simply age discrimination or age alone.

The factors other than age which were associated with a statistically significant increased risk of adverse events in these studies were

  1. functional loss: the inability to perform at least one activity of daily living on admission [27, 30], living in a nursing home [26];

  2. comorbidity and severity of illness: severity of illness on admission [29], reduced level of consciousness on admission [30] diagnosis-related group (DRG) complexity [24], having the admitting diagnosis ‘infection’ [26];

  3. demographic factors: being of Afro-American ethnicity [31], being male [26];

  4. receiving less than adequate quality of care (as judged using the RAND scale) [30];

  5. having a prolonged length of stay in the emergency department [25].

The only factor in all the studies which was found to be associated with significantly fewer adverse events was good documentation of functional status during the admission [30]; this halved the risk of adverse event occurrence in this study.

Outcomes of adverse events

Jahnigen et al.'s study [29] was the only one to describe the effect of adverse events on patient experience, stating that all adverse events were likely to have an effect, ranging from discomfort to death. Both Jahnigen et al. [29] and Ackroyd-Stolarz et al. [25] found that patients with adverse events were likely to be in hospital twice as long as those without.

Only Jahnigen et al. [29] and Gillick et al. [26] investigated the incidence and deleterious effects of interventions employed when adverse events occurred. Jahnigen et al. [29] found that restraint and/or tranquilization were ‘often necessary’ in patients who developed delirium, with subsequent risks of pressure sores, aspiration and self-injury. Gillick et al. [26] found that psychotropic drugs and restraints used for delirium, nasogastric tubes inserted for feeding in undernourished patients and catheters inserted for incontinence were more likely to be used in older people than younger patients. This in turn led to complications such as urinary tract infections, septic shock and fractures in older people, with none occurring in the younger patients who received these interventions. The authors acknowledge that this was a very small sample, but based on evidence in the literature at the time, gave a suggested risk of complications from treatment given as a result of adverse events of 1.0% for young patients, and 5.7% for elderly patients (P < 0.0001).

Several of the studies described the proportions of patients who died during the study period, but only Thomas and Brennan [24] related this to adverse event occurrence; however, there was no significant difference between the rates of death considered to be attributable to adverse events in younger compared with older patients. Only Davis et al. [28] investigated whether there was a link between adverse event occurrence and the need for nursing home placement on discharge from hospital; in this study, most adverse events were minor and transient and their occurrence did not correlate with either nursing home placement or change in level of care on discharge.


Summary of main findings

This review shows that, despite their well-documented vulnerability to adverse events compared with younger people, there have been only a small number of studies specifically designed to look at adverse events in older medical inpatients. This is probably a reflection of the methodological challenges presented by the complexity inherent to these frail patients. Adverse events during a hospital admission that truly occur as a result of the health care process rather than as a result of illness itself are difficult to identify and characterize in patients who have a multitude of pre-existing problems, who present with non-specific presentations of disease, who are less able to communicate or participate in care and who often experience a complex series of health care processes.

In addition, there seems to be variation in the accepted range of adverse events that are relevant to frail older people in hospital. It has been argued that the geriatric syndromes, if they arise during a hospital admission, should be considered to be adverse events because they are (in theory at least) preventable, and all cause harm to patients in terms of morbidity, mortality, length of stay and the need for further care [3235]. However, of the nine studies identified in this systematic review, two were re-analyses of data collected in larger populations, from studies designed without adverse event types in older people in mind, and, therefore, did not explicitly consider the occurrence of the geriatric syndromes as adverse events. This is not just a problem with retrospective case record review studies; other sources of adverse event characterization, such as analysis of data from incident reporting systems, also tend to be skewed towards specific adverse event types that are perhaps more obvious and easily recognizable, such as falls, prescribing errors or pressure sores. The remaining seven studies were designed to look for problems of particular relevance to older people, although each focused on a slightly different set of problems and none provided a comprehensive picture of all of the types of adverse events that older people may experience.

A wide variety of incidences of adverse events were reported in these studies, a finding which reflects the diverse study definitions and aims. All of the studies that compared age groups found statistically significantly more adverse events in older patients than younger ones. This, together with the evidence from data extracted from the large international adverse event studies, provides very strong evidence of an association between age and adverse events. Where explored, this relationship seems to be related to complexity, frailty and co-morbidity rather than age alone. However, sampling strategies used varied depending on the aims of the studies and resulted in varying proportions of frailer, dependent people being included, which limits conclusions about patient safety in these patients.

Limited information was provided about the incidence of clinically relevant outcomes of adverse events such as mortality and institutionalization, perhaps because of the inherent difficulties here in attributing cause and effect. Similarly, the reason as to why the effect of adverse events on length of stay has not been more widely studied is probably that it is very difficult to separate additional days in hospital caused by adverse events from additional days that these frail complex patients may require to recover from functional loss due to illness rather than adverse events. The relationship is further complicated by the fact that any additional length of stay will inevitably result in exposure to further risk of adverse events, a fact acknowledged by Sari et al. [23], who adjusted for length of stay when estimating adverse events.

In addition, limited information is also available regarding preventability of adverse events in older people, presumably because of the confounding effects that co-morbidity and complexity have on making judgements about preventability. However, in theory, the development of new geriatric syndromes should be preventable.

Limitations of this review

The main limitation of this review is the small number and heterogeneity of the studies that it identified, as described above. The wide variation in study quality as judged by the quality assessment scores given above reflects the variation among the studies in terms of their aims and design. Also, this review was not designed to look for the incidences and causal factors of specific conditions such as delirium, falls or incontinence, instead relying on the broad term ‘adverse event’ to include these.

Many of the studies included here are old. Obviously, the ways in which health care is delivered in general, but particularly in geriatric medicine, have changed significantly over the last few decades, which may make the relevance of the results of the earlier studies in this review questionable to practice today.

Conclusions and implications for further work

It is apparent that a great deal of further work to investigate the complex nature of adverse events in older people is required. If we are to improve the safety and quality of care for older people, incorporating measurement approaches tailored to their needs, we need to have clear definitions of the adverse events that affect them, and that must include the geriatric syndromes, problems specific to the care of older people.

Therefore, future studies might assess the scale of adverse events, including all of the geriatric syndromes in large hospital populations. These problems are difficult to identify accurately from case records and are not identifiable in hospital administrative data. This would create the challenge of creating novel methods which would allow accurate distinction between geriatric syndromes present on admission to hospital and those arising as new, preventable adverse events caused by the process of health care during the admission.

Given the profound implications for older people of the acquisition of the geriatric syndromes in hospital, we propose that they should be included as part of routine safety data collection systems in this population. Despite consequent resource implications, the aim of this tailored safety measurement approach would be that common risk factors and underlying causes of the geriatric syndromes could be sought and addressed, leading to improved care for older people in hospitals.


The Centre for Patient Safety and Service Quality is funded by the National Institute for Health Research.


View Abstract