Abstract

Aim: To determine the impact of the introduction of an acute medical admission unit (AMAU) on all-cause hospital mortality in unselected patients undergoing acute medical admission to a teaching hospital.

Design: Analysis of data recorded in the hospital in-patient enquiry (HIPE) system relating to all emergency medical patients admitted to St James's Hospital (SJH), Dublin between 1 January 2002 and 31 December 2006.

Methods: The reference year was 2002, during which patients were admitted to a variety of wards under the care of a named consultant physician. In 2003, two centrally located wards were re-configured to function as an AMAU, and all emergency medical patients were admitted to this unit following emergency department evaluation. Hospital mortality was obtained from a database of deaths occurring during this period and linked to HIPE data.

Results: Following the introduction of the AMAU process, all-cause hospital mortality decreased from 12.6% in 2002 to 7.0% in 2006 (P < 0.0001), representing a 44.4% relative reduction during the course of the 5-year observation period (P < 0.0001). The Odds ratio (95% confidence interval) for all-cause mortality in 2006 compared with 2002 was 0.28 (0.23, 0.35). This effect was powerfully independent of other covariates, including Charlson co-morbidity and illness severity score (APACHE II), in binary logistic regression analysis and was observed across a wide cross-section of diagnostic groups.

Conclusions: The introduction of an AMAU significantly improved all-cause hospital mortality in acute unselected medical patients. The delivery of Acute Medicine may be enhanced by structural reform with emphasis on focus and volume. Prospective studies validating similar models elsewhere should be explored.

Introduction

The concept that focus and volume in healthcare delivery may be associated with superior outcome is not a novel one. Specialist hospital care, provided at high volume centres is now recommended for management of an increasing selection of diseases.1–4 It is the responsibility of senior healthcare managers and clinicians to engineer systems of healthcare delivery that produce measurable improvements in important patient outcomes, including hospital mortality. Few studies to date have examined the impact of structural reforms in the process of hospital care on survival in unselected medical patients.5

This group has previously described the development and use of an acute medical admission unit (AMAU) as a means of providing efficient multidisciplinary care for medical patients requiring emergency hospital admission.6–11 This model has been associated with improvement in a number of key healthcare quality indicators, including cost utilization, length of hospital stay (LOS) and emergency department (ED) waiting times.9–11

This report describes the impact of the AMAU on all-cause hospital mortality in unselected patients undergoing acute medical admission via the ED of a large teaching hospital over the 5-year period 2002–06.

Methods

Background

St James's Hospital (SJH), although a tertiary referral centre for various specialties, operates a daily sectorized acute general medical ‘take-in’ serving as a secondary care centre for emergency medical admissions for its local Dublin catchment area. In 2002, prior to the introduction of the AMAU, emergencies in acute medicine were initially assessed by the staff of the ED and referred by them to the on-call medical team of the day. Any patient requiring emergency hospitalization, (excluding admission to the coronary care unit) was admitted to any available hospital bed, under the care of the ‘on-call’ consultant physician. Fourteen consultant physicians, (all dually accredited in Internal Medicine and a major subspecialty), of whom 10 were whole-time health service consultants and 4 held split service/academic appointments, were responsible for the management of these patients. The ‘on-call’ roster was a 1:9, operated by teams from respiratory medicine, gastroenterology, diabetes/endocrinology, clinical pharmacology, rheumatology and general internal medicine.

Establishment of the AMAU

In 2003 (from 17 March), two of the modern centrally located medical wards, with close proximity to the ED and diagnostic imaging department, were re-configured to function as an AMAU. A detailed operational plan for the unit was devised following extensive discussions with all interested parties in the year prior to its inception. Two general physicians agreed to act as Director and Deputy Directors of the unit, and advise a steering group of physician colleagues on policy matters. Emergencies in acute medicine were assessed as previously by the staff of the ED and referred by them to the ‘on-call’ team of the day. Any patient requiring hospitalization was admitted directly to the AMAU from the ED. The 59-bed AMAU capacity was such that, with an average of 15 admissions each day, up to 70% of all admissions would be predicted to receive their entire hospital care within the unit (maximum permitted stay in AMAU—5 days). Those patients requiring a longer stay were transferred from the AMAU to an appropriate specialty or general medical bed. The principal innovation introduced in this novel system was that all acutely ill medical patients were now admitted from the ED to a single location. Between 2003 and 2006, the on-call roster remained unaltered at 1:9, with each physician on-call for 24 h and a post-call ward round carried out each morning in the AMAU, with other fixed commitments cancelled to accommodate this. Radiology, endoscopy, laboratory services, physiotherapy, occupational therapy, speech and language therapy, clinical nutrition and social services prioritized appropriate requests from the AMAU. All patients identified as suitable for fast-track discharge had a provisional discharge date identified on the post-call ward round. Medical teams reviewed these patients early on the morning of discharge, so that discharge could be confirmed, and arrangements made to transfer the patient to the hospital's discharge lounge in order to vacate beds for patients awaiting admission in the ED. A discharge manager was appointed to help identify patients suitable for early discharge, and work with the multidisciplinary team to ensure timeliness of discharge.

Data collection

A patient database was created by linking the computerized patient administration system (PAS) to the hospital in-patient enquiry (HIPE) scheme.12 HIPE is a national database of coded discharge summaries from acute public hospitals in Ireland.13,,14 A total of 60 hospitals participate in the system and it is an invaluable source of hospital activity level and accreditation.15 Ireland has used the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)16,,17 for both diagnosis and procedure coding from 1990 to 2005, with updates every 5 years. Since 2005, ICD-10-CM has been used. Linking the HIPE dataset with the PAS dataset permits application of routinely collected data for the purposes of research, planning and quality control. Data collected includes hospital number, patient's name, dates of admission and discharge, date of birth, sex, area of residence by county, principal diagnosis, up to nine additional secondary diagnoses, procedures (principal and up to nine additional secondary procedures) and consultant responsible for care. Additional information uploaded to the database included physiological, haematological and biochemical datasets from the ED admission. The HIPE dataset of all coded diseases at time of discharge/death, together with procedures and investigations undertaken during the hospital stay was examined. Codes with <20 occurrences were not considered for analysis. Individual codes together with the combination of all related codes were evaluated. Acute illness severity was quantified using a modified version of the acute physiology and chronic health evaluation (APACHE) II score, calculated with risk scores allocated according to the published method.18 All available APACHE II component variables were used to derive these modified scores (including all but arterial pO2 and pH) in all patients with sufficiently complete data. Data relating to emergency medical patients admitted to SJH between 1 January 2002 and 31 December 2006 were recorded. Data were not recorded prior to 2002.

Statistical methods

Descriptive statistics were calculated for background demographic data, including means/standard deviations (SD), medians/interquartile ranges (IQR), or percentages. Comparisons between categorical variables and mortality were made using chi-square tests. Logistic regression analysis was used to examine the association between year, month, day and time of admission, and hospital mortality, adjusting for the major disease by category (MDC), Charlson co-morbidity index,19 modified APACHE II score, number of admissions, acute or non-acute ward, age and gender. Odds ratios (OR) and 95% confidence intervals (CI) were calculated where appropriate. Statistical significance at P < 0.05 was assumed throughout. JMP statistical software (SAS Institute Inc.) was used for analysis.

Results

Patient characteristics

A total of 33 367 episodes were recorded among 19 528 patients admitted acutely via the ED in the 60 months observation period, between 1 January 2002 and 31 December 2006 (Table 1). The patients included were all medical admissions admitted during the period. No distinction was made between medical admissions admitted to the AMAU, the intensive care unit (ICU)/high dependency unit (HDU), or a small number admitted to other wards for the purpose of analysis.

Table 1

Characteristics of acute medical patients admitted in 2002 and following the establishment of an AMAU 2003–06

 Year of admission 
 2002 2003 2004 2005 2006 P-value* 
 
Episodes 5476 6029 5957 6650 6254  
Patients - last episode 3037 3428 3576 4481 5006  
Patients - any episode in year 4378 4825 4727 5327 5006  
Median (IQR) age, years 62.4 (39.4, 77.1) 62.6 (40.6, 76.9) 64.4 (42.2, 77.4) 62.3 (40.1, 77.2) 62.0 (41.5, 77.8) 0.09 
Sex, male (%) 48.5 48.1 47.8 48.3 47 0.71 
Median (IQR) Charleson comorbidity index, score 0–8a 0 (0,1) 0 (0,1) 0 (0,1) 0 (0,1) 0 (0,1) 0.0001 
Charleson index >0, patients (%) 39.7 35.3 34.7 40.8 46.4 0.001 
Median (IQR) modified APACHE II scoreb 6 (2, 8) 6 (2, 8) 8 (5, 10) 7 (4, 10) 7 (4, 10) <0.0001 
Median (IQR) number of patients in ED awaiting beds at 7 am 14 (8, 19) 9 (5, 13) 8 (4, 14) 12 (8, 16) 2 (0, 13) 0.001 
Median (IQR) length of hospital stay, days 7 (3, 15) 5 (2, 12) 5 (2, 13) 6 (2, 13) 5 (2, 12) 0.0001 
 Year of admission 
 2002 2003 2004 2005 2006 P-value* 
 
Episodes 5476 6029 5957 6650 6254  
Patients - last episode 3037 3428 3576 4481 5006  
Patients - any episode in year 4378 4825 4727 5327 5006  
Median (IQR) age, years 62.4 (39.4, 77.1) 62.6 (40.6, 76.9) 64.4 (42.2, 77.4) 62.3 (40.1, 77.2) 62.0 (41.5, 77.8) 0.09 
Sex, male (%) 48.5 48.1 47.8 48.3 47 0.71 
Median (IQR) Charleson comorbidity index, score 0–8a 0 (0,1) 0 (0,1) 0 (0,1) 0 (0,1) 0 (0,1) 0.0001 
Charleson index >0, patients (%) 39.7 35.3 34.7 40.8 46.4 0.001 
Median (IQR) modified APACHE II scoreb 6 (2, 8) 6 (2, 8) 8 (5, 10) 7 (4, 10) 7 (4, 10) <0.0001 
Median (IQR) number of patients in ED awaiting beds at 7 am 14 (8, 19) 9 (5, 13) 8 (4, 14) 12 (8, 16) 2 (0, 13) 0.001 
Median (IQR) length of hospital stay, days 7 (3, 15) 5 (2, 12) 5 (2, 13) 6 (2, 13) 5 (2, 12) 0.0001 

IQR: interquartile range; ED: emergency department.

*Refers to between-year comparisons; aAnalysis of mean ranks shows an increase over time; bModified APACHE II scores were calculated using available components of the original scoring system (all but arterial pO2 and pH), in all patients (67.8% of total group) with complete available data.

In the reference year prior to the establishment of the AMAU (2002), 3049 patients were admitted over 5476 episodes. The median (IQR) age was 62.4 (39.4, 77.1) years and 48.5% were male. The proportion of patients with any co-morbidity (Charlson co-morbidity score > 0) was 39.7%. Of the total patient group, sufficient data to calculate modified APACHE II scores were available in 67.8%. The median (IQR) modified APACHE II score in the reference year in these patients was 6 (2, 8). The median (IQR) number of admitted acute medical patients in the ED awaiting beds at 7:30 am was 14 (8, 19).

Following establishment of the AMAU, there was an overall increase of 14.2% (5476–6254) in acute medical episodes and of 64.2% (3049–5006) in patients admitted between 2002 and 2006. Median (IQR) daily AMAU admission numbers increased from 14 (11, 17) in 2003 to 17 (14, 21) in 2006 (P < 0.0001, data not shown). Demographic characteristics were compared between years over the 5-year period. Age and sex of the patients did not differ significantly between years. The Charlson co-morbidity index increased significantly over the time period, as did the percentage of patients with any co-morbidity (P < 0.0001 for both). A modest, though highly statistically significant increase in acute illness severity was also observed over time, with the median (IQR) modified APACHE score increasing to 7 (4, 10) by 2006 (P < 0.0001). The MDC mix appeared constant over time for the categories reported (Table 2). Delays in the ED decreased significantly; the median (IQR) daily number awaiting beds in 2006 at 7:30 am was 2 (0, 13).

Table 2

Mortality by major diagnostic category (MDC) in acute medical patients admitted 2002–06

 MDC 
 Respiratory** Nervous* Circulatory* Digestive* Endocrine Musculo skeletal Infectious/ parasitic Other** 
 
2002         
    Number (% within year) 661 (21.7) 528 (17.3) 484 (15.9) 376 (12.4) 98 (3.2) 98 (3.2) 8 (3.2) 713 (23.4) 
    Mortality, number (%) patients 134 (20.3) 63 (11.9) 73 (15.1) 29 (7.7) 5 (5.1) 4 (4.1) 17 (19.8) 58 (8.1) 
2003         
    Number (% within year) 831 (24.3) 572 (16.7) 597 (17.4) 354 (10.3) 94 (2.7) 101 (3.0) 66 (1.9) 807 (23.6) 
    Mortality, number (%) patients 163 (19.6) 45 (8.0) 64 (10.7) 32 (9.0) 1 (10.7) 2 (2.0) 18 (27.3) 76 (9.4) 
2004         
    Number (% within year) 777 (21.7) 537 (15.0) 557 (15.6) 403 (11.3) 97 (2.7) 82 (2.3) 74 (2.1) 1054 (29.4) 
    Mortality, number (%) patients 130 (16.6) 42 (7.8) 55 (9.9) 19 (4.7) 1 (1.0) 0 (0.0) 22 (29.7) 120 (11.4) 
2005         
    Number (% within year) 906 (25.6) 868 (24.0) 788 (22.1) 561 (15.7) 130 (3.6) 157 (4.4) 118 (3.3) 932 (20.9) 
    Mortality, number (%) patients 157 (17.3) 73 (8.4) 88 (11.2) 21 (3.7) 3 (2.3) 3 (1.9) 25 (21.2) 48 (5.2) 
2006         
    Number (% within year) 1173 (23.4) 948 (18.9) 857 (17.1) 553 (11.0) 172 (3.4) 148 (3.0) 68 (1.4) 1087 (21.7) 
    Mortality, number (%) patients 119 (10.1) 59 (6.2) 74 (8.6) 19 (3.4) 5 (3.4) 0 (0.0) 15 (22.1) 59 (5.4) 
 MDC 
 Respiratory** Nervous* Circulatory* Digestive* Endocrine Musculo skeletal Infectious/ parasitic Other** 
 
2002         
    Number (% within year) 661 (21.7) 528 (17.3) 484 (15.9) 376 (12.4) 98 (3.2) 98 (3.2) 8 (3.2) 713 (23.4) 
    Mortality, number (%) patients 134 (20.3) 63 (11.9) 73 (15.1) 29 (7.7) 5 (5.1) 4 (4.1) 17 (19.8) 58 (8.1) 
2003         
    Number (% within year) 831 (24.3) 572 (16.7) 597 (17.4) 354 (10.3) 94 (2.7) 101 (3.0) 66 (1.9) 807 (23.6) 
    Mortality, number (%) patients 163 (19.6) 45 (8.0) 64 (10.7) 32 (9.0) 1 (10.7) 2 (2.0) 18 (27.3) 76 (9.4) 
2004         
    Number (% within year) 777 (21.7) 537 (15.0) 557 (15.6) 403 (11.3) 97 (2.7) 82 (2.3) 74 (2.1) 1054 (29.4) 
    Mortality, number (%) patients 130 (16.6) 42 (7.8) 55 (9.9) 19 (4.7) 1 (1.0) 0 (0.0) 22 (29.7) 120 (11.4) 
2005         
    Number (% within year) 906 (25.6) 868 (24.0) 788 (22.1) 561 (15.7) 130 (3.6) 157 (4.4) 118 (3.3) 932 (20.9) 
    Mortality, number (%) patients 157 (17.3) 73 (8.4) 88 (11.2) 21 (3.7) 3 (2.3) 3 (1.9) 25 (21.2) 48 (5.2) 
2006         
    Number (% within year) 1173 (23.4) 948 (18.9) 857 (17.1) 553 (11.0) 172 (3.4) 148 (3.0) 68 (1.4) 1087 (21.7) 
    Mortality, number (%) patients 119 (10.1) 59 (6.2) 74 (8.6) 19 (3.4) 5 (3.4) 0 (0.0) 15 (22.1) 59 (5.4) 

*P < 0.01, **P < 0.001 for change in mortality from reference year; Mortality rates were calculated as percentage of individual patients admitted, rather than percentage of admission episodes.

Mortality

Of the total patient group, 1947 deaths occurred during the 5-year observation period. These comprised all deaths occurring at any time during a hospital episode, under the care of any medical admitting team. Following the introduction of the AMAU, all-cause hospital mortality fell significantly during 2003–06 (Figure 1A and B). The annual mortality rate in acute medical patients decreased from 12.6% in 2002 to 7.0% in 2006 (Figure 1A), representing a 44.4% relative reduction during the course of the study (P < 0.0001). As the AMAU system was felt unlikely to directly influence mortality in patients whose hospital stay was prolonged, 30-day mortality was also examined in an effort to control for varying numbers of long-stay patients occupying acute hospital beds (Figure 1B). A similar, highly statistically significant decrease in mortality over time was observed, largely beginning in 2004. Falling mortality was also observed consistently after 2002 across the majority of individual MDC's, containing some 90% of the total patient group (Table 2). While large relative decreases in the small mortality rates occurring in the endocrine and musculoskeletal MDCs were observed, these changes did not reach statistical significance. Mortality did not change during the study in patients classified in the infectious/parasitic disease category. The median patient age in this subgroup (43.4 years) was substantially lower than in any other MDC and 87.8% of patients had specific diagnoses of septicaemia or sepsis syndrome (data not shown). Adjusted ORs for factors independently associated with risk of death during the study are presented in Table 3. The adjusted OR (95% CI) for all cause in-patient mortality in 2006 compared with 2002 was 0.28 (0.23–0.35).

Figure 1.

A Annual and B 30-day all-cause hospital mortality rates in acute medical patients admitted in 2002 (reference year, white bar) and following the introduction of an acute medical admission unit from 2003 to 2006 (grey bars). Whiskers indicate 95% confidence intervals. See text for further definitions.

Figure 1.

A Annual and B 30-day all-cause hospital mortality rates in acute medical patients admitted in 2002 (reference year, white bar) and following the introduction of an acute medical admission unit from 2003 to 2006 (grey bars). Whiskers indicate 95% confidence intervals. See text for further definitions.

Table 3

Logistic regression predicting in-hospital death (vs. survival) in acute medical patients admitted between 2002 and 2006

 Odds ratio Lower CI Upper CI P-value 
 
Male vs female 1.25 1.12 1.38 <0.0001 
Age on admission 211.99 151.81 296.04 <0.0001 
Year of admission    <0.0001 
    2003 vs 2002 1.81 1.47 2.22 <0.0001 
    2004 vs 2002 1.2 0.98 1.48  
    2005 vs 2002 0.73 0.6 0.9  
    2006 vs 2002 0.28 0.23 0.35  
Month of admission    0.004 
    Jan vs Aug 1.56 1.15 2.12  
    Feb vs Aug 1.1 0.77 1.55  
    Mar vs Aug 1.21 0.87 1.68  
    Apr vs Aug 1.42 1.01 1.97  
    May vs Aug 0.7 1.39  
    Jun vs Aug 1.11 0.78 1.55  
    Jul vs Aug 0.71 0.49 1.01  
    Sep vs Aug 0.84 0.59 1.17  
    Oct vs Aug 1.08 0.77 1.5  
    Nov vs Aug 0.98 0.7 1.35  
    Dec vs Aug 0.96 0.69 1.33  
Time of admission    0.001 
    08:00–16:00 vs 00:00–08:00 0.98 0.84 1.13  
    16:00–24:00 vs 00:00–08:00 1.28 1.12 1.48  
Acute ward vs not acute ward 1.83 1.65 2.03 <0.0001 
Major diagnostic category*    <0.0001 
    Respiratory system 3.04 5.35  
    Nervous system 1.27 0.92 1.75  
    Circulatory system 1.48 1.09 2.03  
    Digestive system 0.59 0.4 0.9  
    Endocrine, nutritional, metabolic 0.12 0.04 0.29  
    Infectious, parasitic disease 37.14 21.9 63.01  
    Others 1.27 0.92 1.75  
Charlson comorbidity index 9.27 6.53 13.14 <0.0001 
Number of admissions per patient    <0.0001 
    ⩾3 vs 1 1.84 1.52 2.23  
    2 vs 1 0.96 0.8 1.16  
 Odds ratio Lower CI Upper CI P-value 
 
Male vs female 1.25 1.12 1.38 <0.0001 
Age on admission 211.99 151.81 296.04 <0.0001 
Year of admission    <0.0001 
    2003 vs 2002 1.81 1.47 2.22 <0.0001 
    2004 vs 2002 1.2 0.98 1.48  
    2005 vs 2002 0.73 0.6 0.9  
    2006 vs 2002 0.28 0.23 0.35  
Month of admission    0.004 
    Jan vs Aug 1.56 1.15 2.12  
    Feb vs Aug 1.1 0.77 1.55  
    Mar vs Aug 1.21 0.87 1.68  
    Apr vs Aug 1.42 1.01 1.97  
    May vs Aug 0.7 1.39  
    Jun vs Aug 1.11 0.78 1.55  
    Jul vs Aug 0.71 0.49 1.01  
    Sep vs Aug 0.84 0.59 1.17  
    Oct vs Aug 1.08 0.77 1.5  
    Nov vs Aug 0.98 0.7 1.35  
    Dec vs Aug 0.96 0.69 1.33  
Time of admission    0.001 
    08:00–16:00 vs 00:00–08:00 0.98 0.84 1.13  
    16:00–24:00 vs 00:00–08:00 1.28 1.12 1.48  
Acute ward vs not acute ward 1.83 1.65 2.03 <0.0001 
Major diagnostic category*    <0.0001 
    Respiratory system 3.04 5.35  
    Nervous system 1.27 0.92 1.75  
    Circulatory system 1.48 1.09 2.03  
    Digestive system 0.59 0.4 0.9  
    Endocrine, nutritional, metabolic 0.12 0.04 0.29  
    Infectious, parasitic disease 37.14 21.9 63.01  
    Others 1.27 0.92 1.75  
Charlson comorbidity index 9.27 6.53 13.14 <0.0001 
Number of admissions per patient    <0.0001 
    ⩾3 vs 1 1.84 1.52 2.23  
    2 vs 1 0.96 0.8 1.16  

Higher odds ratios indicate a greater likelihood of death; CI: confidence interval.

*Reference for ORs = musculoskeletal system.

Logistic regression was performed to examine other factors associated with hospital mortality. These results are also presented in Table 3. Significant independent predictors of death included advanced age (P < 0.0001), male gender [OR (95% CI) 1.25 (1.12–1.38)], MDC (P < 0.0001), month of admission (January and August associated with highest and lowest mortality, respectively, P = 0.0006), time of day admitted: 4 pm—midnight vs. midnight-8 am OR (95% CI) 1.28 (1.12–1.48), acute vs. non-acute ward [OR (95% CI) 1.83 (1.65–2.03)] and Charlson co-morbidity index (P < 0.0001). Two recorded admissions did not predict an adverse outcome, but three or more was significantly associated with increased mortality, with an OR (95% CI) of 1.84 (1.52–2.23) compared with one admission only. Of note, factors such as day of the week admitted, numbers of patients in the ED awaiting beds at 7 am, and numbers of patients admitted during a 24 h on-call period did not significantly influence mortality in this analysis (data not shown). Modified APACHE II score was also evaluated as a predictor of mortality. As expected, the APACHE II score was strongly associated with annual mortality in univariate analysis (Figure 2), and when included in multiple regression analysis employing the same design as presented in Table 3, independently predicted death with an adjusted OR (95% CI) of 1.13 (1.11, 1.15) (data not shown). Importantly, even controlling for these expected predictors of adverse outcome, mortality remained independently and strongly associated with the structural changes introduced over the period 2003–06 (P < 0.0001).

Figure 2.

Annual all-cause hospital mortality rates in acute medical patients admitted between 2002 and 2006 arranged according to deciles of modified APACHE II score. Modified APACHE II scores were calculated using available components of the original scoring system (all but arterial pO2 and pH), in all patients with complete available data (67.8% of total group).

Figure 2.

Annual all-cause hospital mortality rates in acute medical patients admitted between 2002 and 2006 arranged according to deciles of modified APACHE II score. Modified APACHE II scores were calculated using available components of the original scoring system (all but arterial pO2 and pH), in all patients with complete available data (67.8% of total group).

Discussion

In this article, describing the outcome for 19 528 unselected acute medical patients admitted via the ED of a busy teaching hospital, the development of a dedicated AMAU was associated with 45 and 36% relative reductions in all-cause annual and 30-day hospital mortality, respectively, despite significant increases in workload, co-morbidity and acute illness severity over time. This effect was powerfully independent of other covariates in binary logistic regression analysis and was observed across a wide cross-section of diagnostic groups. Additional independent predictors of mortality included male sex, advanced age, month of admission, admission between 4 pm and midnight, MDC, high Charlson co-morbidity index, modified APACHE II score and >2 previous admissions. Finally, the establishment of an AMAU was associated with significantly reduced delays in transfer of patients from the ED to medical wards and lengths of hospital stay.

Our description of structural reform in acute medicine is observational, uncontrolled and therefore suffers unknown bias. It can never have the rigour of a randomized trial. Our intent however, based on this experience, is to challenge current assumptions that the mode of care delivery in acute medicine is not an important issue. This is the first prospective study to demonstrate substantial and statistically significant improvement in all-cause hospital mortality associated with the use of such a process. The effect size was very large. The numbers to treat (NNT) using the novel system to save one life comparing 2002 and 2006 was 18.0.

The unit permits provision of care at a single convenient location within the hospital during the critical hours immediately following acute admission. The admission ward, a focal point of acute activity, has encouraged increased availability and responsiveness of senior decision makers. As a result, care can be delivered in a manner that is both expert and efficient in terms of doctors’ and health professionals’ time and energy. Unproductive periods spent travelling within the hospital campus are minimized. In a recent study evaluating work practice reform in over 550 American teaching hospitals, the introduction of improved work hours for medical staff was associated with improved hospital mortality in medical patients, strongly suggesting that fatigue in hospital staff adversely effects patient care.20

In this model, the AMAU is an acute medical facility where <70% of patients will spend all of their hospital admission. This is possible because of the relationship between the median daily ‘take’ of 17 and the total bed number of 59. The nursing staff and allied health professionals who are part of the AMAU are exclusively engaged in the care of acutely ill medical patients. The AMAU is not, unlike other models, a way station to facilitate specialty triage5—in this respect, the term ‘admission’ might best be omitted from its title. This system is not based on the assumption that the majority of patients can or should be triaged to an appropriate specialty; rather it presumes most patients, with ageing, will have multiple co-morbidities. The patients are admitted under a rotating Consultant Physician of the day—an experienced specialist, who in addition to a major medical subspecialty, is dually certified in General (Internal) Medicine. The patients, by and large, remain under the care of the admitting consultant and this receiving physician is advised by the different specialties; the challenge for the consultant and the team is to apply and integrate the advice. It can be argued that constant feedback from colleagues maintains essential skills. In this model continuity of care has to date been maintained. In Ireland there has been little dispensation to the European Working time directive and consequently shift work patterns have been unusual. Ward rounds can anticipate the presence of all team members on most occasions; all essential knowledge relevant to clinical decisions should therefore be available.

The AMAU provides focused systems of acute medical care for large volumes of patients. In the present study, it is hypothesized that the mortality improvements observed between 2002 and 2006 occurred as a consequence of the systematic structural reforms that were introduced during the period. In one previous study evaluating the early distribution of acute medical patients to appropriate specialist care, with interim care provided in an acute medicine unit established in 1999, some of the mortality benefit that followed the unit's introduction simply reflected a pre-existing underlying downward mortality trend.5 Since mortality data in acute medical admissions were not systematically recorded for years prior to 2002 in the present study, a trend of this kind cannot be definitively discounted as a confounder. However, It appeared to us that a major change in outcome did not commence for at least 1 year from the AMAU's inception: 30-day all-cause mortality began to decrease only in 2004, a year after the novel system had been established. The increase over time in the Apache II score observed in the present analysis would also be inconsistent with such an hypothesis. Moreover, the concept that increased specialization and volume produces improved healthcare outcomes has been applied and validated in numerous other clinical settings. For example, care in dedicated specialist units has been consistently associated with improved morbidity, mortality and LOS in patients suffering from stroke.4,,21 Individual studies and meta-analyses in broader areas such as cancer care,1 acute coronary intervention,2 and in many surgical sub-specialties3 have also demonstrated that highly specialized units or individuals dealing with large numbers of similar patients tend to produce superior results across a range of outcomes, including mortality. Taken together, these observations suggest that focus and volume produce superior outcomes in a generic way that is largely independent of disease settings. Consistent with this concept, mortality improvements were observed across a wide range of diagnostic categories in this analysis. A number of specialty-specific developments took place in the hospital between 2002 and 2006, such as the development of non-invasive ventilation (NIV) services, that would be expected to influence mortality in specific diagnostic groups.22,,23 The broad mortality benefit attributable to year of admission however, was independent of MDC in regression analysis. Notably, no improvement in mortality was seen in patients classified in the ‘infectious/parasitic’ MDC. This small group, comprising some 2–4% of patients, consisted almost entirely of patients with a diagnosis of either septicaemia or sepsis syndrome, and appears to represent a particularly ill cohort of patients, with high predicted mortality irrespective of intervention.24

A number of additional independent mortality predictors were identified during the period of observation. Age has predicted mortality in previous studies evaluating emergency medical admissions25 and in the ICU setting.26 Gender also predicted death in the present study, with a male to female mortality OR of 1.25. Life expectancies are lower in males than females in most developed countries.27 The influence of gender on hospital mortality, however, varies considerably between previous studies.28–30 The effect of month of admission on death rates likely reflects more severe disease occurring in winter months, as the highest and lowest mortality were observed in January and August, respectively. It has been hypothesized that the arrival of new and inexperienced medical staff in acute hospitals in July may produce adverse outcomes.31,,32 Our observations in contrast suggest that any ‘July effect’ in acute medical patients relates an improved outcome to seasonality. A further novel aspect of the article concerns the association between time of admission and hospital mortality. Admission between 4 pm and midnight was associated with significantly increased mortality. While ‘out of hours’ admission has previously been linked to increased LOS in emergency medical care,33 our data highlight nighttime admission as an independent determinant of increased mortality. This may be due to differences either in the severity/type of illness in patents presenting to the ED during the night period or in access to emergency diagnostic and therapeutic systems.34

As expected, patients with higher levels of co-morbid disease, higher APACHE II scores, and those requiring frequent admission experienced higher mortality rates. The prognostic value of the Charlson index and APACHE II scores have been established,18,,19 and repeated hospitalization has been adversely associated with survival in previous studies involving medical patients.35,,36 Frequent re-admission was associated with increased LOS in unselected medical patients in a previous study conducted at this centre.7

This group has now demonstrated over several studies that the introduction of an AMAU as a focused system for managing acute medical patients in a large teaching hospital has been associated with dramatic improvements in numerous healthcare quality indicators, now including all-cause hospital mortality.9,,10 While the present data are observational and lack the rigour of a randomized controlled trial, the design of the study was deliberately hypothesis-generating and not as a means to definitive proof-of-concept. The magnitude of the observed effects challenge current assumptions that the process of care for acute medical emergencies is not an important issue. Our observations support the hypothesis that the observed improvements in mortality and other relevant outcomes are attributable to the structured care processes introduced with the advent of the AMAU. Prospective studies validating similar models elsewhere should be explored.

Acknowledgements

We wish to recognize the contribution of our consultant medical colleagues and the non-consultant members of the ‘on-call’ teams without which this initiative could not have progressed. In particular their continued enthusiasm for the practice of acute medicine in difficult circumstances within stringent resource constraints deserves recognition. The dedicated contribution of Sr. S. Donnelly, her Clinical Nurse Managers and the ancillary professions related to medicine (SCOPE) is gratefully acknowledged. No external funding or sponsorship was obtained for this study.

Conflict of interest: None declared.

References

1
Hillner
BE
Smith
TJ
Desch
CE
Hospital and physician volume or specialization and outcomes in cancer treatment: importance in quality of cancer care
J Clin Oncol
 , 
2000
, vol. 
18
 (pg. 
2327
-
40
)
2
Nallamothu
BK
Wang
Y
Magid
DJ
McNamara
RL
Herrin
J
Bradley
EH
, et al.  . 
Relation between hospital specialization with primary percutaneous coronary intervention and clinical outcomes in ST-segment elevation myocardial infarction: National Registry of Myocardial Infarction-4 analysis
Circulation
 , 
2006
, vol. 
113
 (pg. 
222
-
9
)
3
Chowdhury
MM
Dagash
H
Pierro
A
A systematic review of the impact of volume of surgery and specialization on patient outcome
Br J Surg
 , 
2007
, vol. 
94
 (pg. 
145
-
61
)
4
Collaborative systematic review of the randomised trials of organised inpatient (stroke unit) care after stroke. Stroke Unit Trialists’ Collaboration
Br Med J
 , 
1997
, vol. 
314
 (pg. 
1151
-
9
)
5
Moore
S
Gemmell
I
Almond
S
Buchan
I
Osman
I
Glover
A
, et al.  . 
Impact of specialist care on clinical outcomes for medical emergencies
Clin Med
 , 
2006
, vol. 
6
 (pg. 
286
-
93
)
6
Moloney
ED
Bennett
K
Silke
B
Length of hospital stay, diagnoses and pattern of investigation following emergency admission to an Irish teaching hospital
Ir Med J
 , 
2004
, vol. 
97
 (pg. 
170
-
2
)
7
Moloney
ED
Bennett
K
Silke
B
Patient and disease profile of emergency medical readmissions to an Irish teaching hospital
Postgrad Med J
 , 
2004
, vol. 
80
 (pg. 
470
-
4
)
8
Moloney
ED
Smith
D
Bennett
K
O’Riordan
D
Silke
B
Do consultants differ? Inferences drawn from hospital in-patient enquiry (HIPE) discharge coding at an Irish teaching hospital
Postgrad Med J
 , 
2005
, vol. 
81
 (pg. 
327
-
32
)
9
Moloney
ED
Smith
D
Bennett
K
O’Riordan
D
Silke
B
Impact of an acute medical admission unit on length of hospital stay, and emergency department ‘wait times’
QJM
 , 
2005
, vol. 
98
 (pg. 
283
-
9
)
10
Moloney
ED
Bennett
K
O’Riordan
D
Silke
B
Emergency department census of patients awaiting admission following reorganisation of an admissions process
Emerg Med J
 , 
2006
, vol. 
23
 (pg. 
363
-
7
)
11
Moloney
ED
Bennett
K
Silke
B
Factors influencing the costs of emergency medical admissions to an Irish teaching hospital
Eur J Health Econ
 , 
2006
, vol. 
7
 (pg. 
123
-
8
)
12
O’Neill
B
The hospital in-patient enquiry scheme-a study of data accuracy
Ir Med J
 , 
1982
, vol. 
75
 (pg. 
238
-
9
)
13
O’Neill
B
The hospital in-patient enquiry scheme - a study of data accuracy
Ir Med J
 , 
1982
, vol. 
75
 (pg. 
238
-
9
)
14
Mehanni
M
Loughman
E
Allwright
S
Prichard
J
The hospital in-patient enquiry scheme: a study of data accuracy and capture
Ir Med J
 , 
1995
, vol. 
88
 (pg. 
24
-
6
)
15
Mehanni
M
Loughman
E
Allwright
SP
Prichard
J
The hospital in-patient enquiry scheme: a study of data accuracy and capture
Ir Med J
 , 
1995
, vol. 
88
 (pg. 
24
-
6
)
16
Slee
VN
The International classification of diseases: ninth revision (ICD-9)
Ann Intern Med
 , 
1978
, vol. 
88
 (pg. 
424
-
6
)
17
MacIntyre
C
Ackland
M
Chandraraj
E
Pilla
J
Accuracy of ICD-9-CM codes in hospital morbidity data, Victoria: implications for publin health research
Aust N Z J Public Health
 , 
1997
, vol. 
21
 (pg. 
477
-
82
)
18
Knaus
WA
Draper
EA
Wagner
DP
Zimmerman
JE
APACHE II: a severity of disease classification system
Crit Care Med
 , 
1985
, vol. 
13
 (pg. 
818
-
29
)
19
Charlson
ME
Pompei
P
Ales
KL
MacKenzie
CR
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
J Chronic Dis
 , 
1987
, vol. 
40
 (pg. 
373
-
83
)
20
Shetty
KD
Bhattacharya
J
Changes in hospital mortality associated with residency work-hour regulations
Ann Intern Med
 , 
2007
, vol. 
147
 (pg. 
73
-
80
)
21
Foley
N
Salter
K
Teasell
R
Specialized stroke services: a meta-analysis comparing three models of care
Cerebrovasc Dis
 , 
2007
, vol. 
23
 (pg. 
194
-
202
)
22
Ram
FS
Picot
J
Lightowler
J
Wedzicha
JA
Non-invasive positive pressure ventilation for treatment of respiratory failure due to exacerbations of chronic obstructive pulmonary disease
Cochrane Database Syst Rev
 , 
2004
(1)
pg. 
CD004104
 
23
Winck
JC
Azevedo
LF
Costa-Pereira
A
Antonelli
M
Wyatt
JC
Efficacy and safety of non-invasive ventilation in the treatment of acute cardiogenic pulmonary edema—a systematic review and meta-analysis
Crit Care
 , 
2006
, vol. 
10
 pg. 
R69
 
24
Cook
DA
Methods to assess performance of models estimating risk of death in intensive care patients: a review
Anaesth Intensive Care
 , 
2006
, vol. 
34
 (pg. 
164
-
75
)
25
Goodacre
S
Turner
J
Nicholl
J
Prediction of mortality among emergency medical admissions
Emerg Med J
 , 
2006
, vol. 
23
 (pg. 
372
-
5
)
26
Venker
J
Miedema
M
Strack van Schijndel
RJ
Girbes
AR
Groeneveld
AB
Long-term outcome after 60 days of intensive care
Anaesthesia
 , 
2005
, vol. 
60
 (pg. 
541
-
6
)
27
Hoyert
DL
Heron
MP
Murphy
SL
Kung
HC
Deaths: final data for 2003
Natl Vital Stat Rep
 , 
2006
, vol. 
54
 (pg. 
1
-
120
)
28
Gordon
HS
Rosenthal
GE
The relationship of gender and in-hospital death: increased risk of death in men
Med Care
 , 
1999
, vol. 
37
 (pg. 
318
-
24
)
29
Galvao
M
Kalman
J
DeMarco
T
Fonarow
GC
Galvin
C
Ghali
JK
, et al.  . 
Gender differences in in-hospital management and outcomes in patients with decompensated heart failure: analysis from the Acute Decompensated heart failure national registry (ADHERE)
J Card Fail
 , 
2006
, vol. 
12
 (pg. 
100
-
7
)
30
Kollef
MH
O’Brien
JD
Silver
P
The impact of gender on outcome from mechanical ventilation
Chest
 , 
1997
, vol. 
111
 (pg. 
434
-
41
)
31
Finkielman
JD
Morales
J
Peters
SG
Keegan
MT
Ensminger
SA
Lymp
JF
, et al.  . 
Mortality rate and length of stay of patients admitted to the intensive care unit in July
Crit Care Med
 , 
2004
, vol. 
32
 (pg. 
1161
-
5
)
32
Barry
WA
Rosenthal
GE
Is there a July phenomenon? The effect of July admission on intensive care mortality and length of stay in teaching hospitals
J Gen Intern Med
 , 
2003
, vol. 
18
 (pg. 
639
-
45
)
33
Earnest
A
Chen
MI
Seow
E
Exploring if day and time of admission is associated with average length of stay among inpatients from a tertiary hospital in Singapore: an analytic study based on routine admission data
BMC Health Serv Res
 , 
2006
, vol. 
6
 pg. 
6
 
34
Bell
CM
Redelmeier
DA
Waiting for urgent procedures on the weekend among emergently hospitalized patients
Am J Med
 , 
2004
, vol. 
117
 (pg. 
175
-
81
)
35
Almagro
P
Calbo
E
Ochoa de Echaguen
A
Barreiro
B
Quintana
S
Heredia
JL
, et al.  . 
Mortality after hospitalization for COPD
Chest
 , 
2002
, vol. 
121
 (pg. 
1441
-
8
)
36
Yohannes
AM
Baldwin
RC
Connolly
MJ
Predictors of 1-year mortality in patients discharged from hospital following acute exacerbation of chronic obstructive pulmonary disease
Age Ageing
 , 
2005
, vol. 
34
 (pg. 
491
-
6
)