International Journal for Quality in Health Care 14:285-293 (2002)
© 2002 International Society for Quality in Health Care
Paper |
Hospital in the home is cost saving for appropriately selected patients: a comparison with in-hospital care
1National Centre for Immunisation Research & Surveillance of Vaccine Preventable Diseases, New Children’s Hospital, Westmead, New South Wales
2Department of Community Health, Melbourne Health, Victoria
3Department of Epidemiology and Preventive Medicine, Monash University, Victoria, Australia
Background. As the cost of acute care in hospitals increases, there is an increasing need to find alternative means of providing acute care. Hospital in the home (HITH) has developed in response to this challenge. Current evidence is conflicting as to whether HITH provides cost savings compared with in-hospital care (IHC). The heterogeneous nature of HITH and the clinical complexity of patients is the greatest obstacle to making valid comparisons between the two modes of care.
Objective. To compare costs and outcomes of HITH to IHC in hospitals in Victoria, Australia.
Data sources/study setting. Hospital morbidity data and medical records from Victoria, Australia.
Study design. A costing study of 924 randomly selected episodes of HITH care, individually matched to 924 comparable IHC episodes.
Methods. Unadjusted total episode costs (TEC) and averaged daily costs for HITH and IHC were calculated. Mortality and length of stay (LOS) were compared for HITH and IHC episodes. Simple linear and multiple regression were used to analyse costing data, while logistic regression was used to compare in-hospital mortality and LOS in HITH versus IHC episodes.
Principal findings. The 1848 episodes of care in the sample represented a heterogeneous range of acute conditions in 31 Victorian hospitals. HITH consisted of two distinct subgroups: pure-HITH (total episode substitution) and mixed-HITH (partial episode substitution). The cost of episodes of acute care containing a HITH component were overall 9% less expensive than IHC (P = 0.04), while pure-HITH was 38% cheaper than matched IHC (P < 0.001). The variable HITH, along with LOS and chemotherapy, explained the 60% variation in TEC. The mean cost of pure-HITH episodes was 22% lower than mixed-HITH for total HITH cost (P = 0.004). The in-hospital mortality rate in HITH (3.8%) and IHC (5.2%) was not significantly different. Pure-HITH was associated with shorter LOS, whereas HITH (mixed and pure) was strongly associated with longer LOS.
Conclusion. In our study the adjusted cost of HITH was significantly cheaper than IHC, particularly as total episode substitution. The cost needs to be adjusted because many factors other than HITH or IHC can influence crude costs. There may be potential for wider use of HITH for appropriately selected patients.
Keywords: acute care, cost, hospital, hospital in the home, outcomes
Hospital in the home (HITH) is a setting in which selected types of health care can be delivered in the home to suitable, consenting patients who require in-patient hospital care [1,2]. Increasing fiscal pressure on hospitals, as well as patient choice, has driven the development of HITH as a viable alternative for provision of acute care, both in Australia and elsewhere [3,4]. In Victoria, the second most populous state in Australia, HITH substitutes for an acute hospital admission and delivers a wide range of acute services to patients with diverse clinical conditions [5,6].
Several studies, including some that used a randomized, controlled design, have shown that patient outcomes are equivalent for patients using HITH and in-hospital care (IHC) [7–10]. Studies have shown that both perceived quality of life and patient satisfaction are greater in HITH than in IHC [8,9,11–13].
However, the cost-effectiveness of HITH care is still debated. This is partly because HITH is not a homogeneous entity, and evaluation studies have focused on different service models, conditions, and treatments, ranging from acute to post-acute and rehabilitation [13,14]. Some studies have shown that HITH does not reduce total costs [15]. One study determined that although HITH is cheaper per day, it results in a longer length of stay (LOS), so that the total cost of HITH is higher [16]. Other studies have shown that the cost of HITH is equivalent to [5] or less than IHC [4,9,17–19]. The results of these studies, however, are not necessarily comparable, as they were conducted on diverse patient groups, such as orthopaedic patients, terminally ill patients, patients with cancer, a variety of infections, or suspected myocardial infarction. In addition, randomized, controlled clinical trial (RCT) methodology has mostly been used to study homogeneous patient groups [8–10,20]. Some RCTs have studied heterogeneous groups, but have often been limited to selected conditions or treatments, or have not been large enough to address the diversity of HITH [20–22]. HITH is a diverse entity, and varies in definition from state to state and country to country. As such, it is difficult to apply the findings of any particular study to HITH as a general form of care.
The limitations of the validity and generalizability of previous costing studies have been identified by several authors [3,23–25]. These studies displayed inconsistencies in cost methods between HITH and institutional costs. Prospective studies commonly measured actual direct inputs for HITH episodes and compared these with average direct inputs for control episodes [4,18].
One of the major limitations of costing studies is the failure to identify and adjust for potential confounding variables, such as severity of patient illness and co-morbidity, which may independently impact on cost. Despite best efforts to match cases of HITH to IHC, these factors may be unevenly distributed between HITH and IHC, and can predict cost independently of HITH/IHC. This means that a simple comparison of the crude cost of HITH versus IHC episodes may be invalid. Without identifying and adjusting for confounding factors, the true cost comparison of HITH versus IHC episodes cannot accurately be determined. Many studies also fail to consider the heterogeneity of HITH and ignore the significant proportion of patients who have less common services and diagnoses.
The aims of this study were to calculate costs of various types of HITH and IHC, to compare costs of HITH with IHC, after accounting for variance associated with age, gender, co-morbidity, and other confounders, and to compare outcomes (mortality and LOS) in HITH versus IHC, after adjusting for potential confounders.
Materials and methods
Study design
This was a costing study comparing matched episodes of HITH with IHC. A retrospective cohort of HITH episodes was initially selected and matched to equivalent in-hospital episodes. The major outcome of interest was cost. Costs were calculated and crude costs were adjusted for potential clinical confounders using logistic regression analysis.
Sample selection and matching
A sample of 1600 consecutive HITH episodes between July and December 1997 was selected from the Victorian hospital morbidity dataset. A computerized algorithm was used to match these to IHC episodes. Matching was carried out to ensure validity in the comparison of cost between HITH and IHC episodes. After first-pass matching, re-matching of unmatched episodes was carried out using a more sensitive algorithm. Episodes were matched initially by: hospital code, age (within 5 years), sex, principal diagnosis, number and type of co-morbidity, number of diagnoses, and date of episode (admission date within 3 months). The second pass matching used: hospital code, age (within 10 years), sex, Australian National Diagnosis-Related Group (ANDRG) [26], and episode date. Computer matching was successful for 1533 HITH episodes from 31 hospitals.
The adequacy of matching was then validated by medical record review, checking all the criteria for matching listed above, and using a list of pre-defined criteria to determine the functional status of the patient, the intensity of care received, and to confirm suitability for HITH. Of the 1533 HITH episodes, 977 were adequately matched to IHC episodes by record review. The sample therefore consisted of 977 HITH and 977 IHC episodes. Of these, 53 pairs (106 episodes) were excluded from cost analysis due to the inability of four hospitals to supply the required costing data. The final sample available for analysis was 924 matched pairs of HITH and IHC (1848 episodes in total).
Costing
Costing was performed using clinical costing systems (TransitionTM, Casemix Information Delivery Service, or TrendstarTM) where available (60% of the sample), or manual costing. The costing method used at the manually costed sites mimicked the TransitionTM costing methodology. Eleven service types were used for cost analysis: nursing, medical, allied health, surgical, theatre, imaging, pathology, pharmacy, emergency, intensive care, and coronary care. An activity-based costing methodology was used to derive unit costs. This involved extracting total operational expenditure from the hospital general ledger, allocating direct and indirect costs to departments, defining cost structures for intermediate products, and applying formulae to the cost structure defined for intermediate products to determine unit costs. Job order costing methodology was then used to derive total episode costs (TEC) from the sum of unit costs for all intermediate products consumed in the patient episode. Averaged daily cost (ADC) was calculated from TEC divided by the LOS.
Statistical methods
The HITH sample was analysed for the subgroups ‘pure-HITH’ (total episode substitution, where the entire episode of care took place in HITH) and ‘mixed-HITH’ (partial episode substitution, where part of the episode of care took place in-hospital and part in HITH), and then compared with their matched IHC episodes. The analysis was performed this way because the pure and mixed HITH groups were very different demographically, clinically and in cost. For example, 100% of chemotherapy episodes in the sample were pure-HITH, whereas 100% of cardiac surgical and cardiac medical episodes treated with HITH were mixed-HITH. LOS in pure-HITH (mean 3.4 days, median 1 day) was significantly lower than for mixed-HITH episodes (mean 14.7 days, median 8 days; P < 0.0001). Only 37% of the LOS for mixed-HITH episodes was days spent in HITH (mean 5.4 days, median 3 days), but this was still significantly longer than the mean LOS for pure HITH (P < 0.0001). Over 25% of mixed-HITH episodes had a LOS >2 standard deviations beyond the mean for the diagnosis-related group (DRG), compared with 5% in IHC and pure-HITH.
Unadjusted mean and median values were calculated for total episode cost, LOS, and ADC for HITH and matched IHC episodes. When mixed-HITH episodes were compared with matched in-hospital episodes, the cost of the HITH component was compared with the average cost of the appropriate days of care in the matched episode, not with that of the entire in-hospital episode.
Data for the entire sample, including HITH and IHC episodes, were analysed for predictors of total cost. Data for all HITH episodes were analysed for predictors of HITH cost.
Service delivery models were defined as either a hospital, general practitioner (GP), or mixed model of care on the basis of who was responsible for medical management of HITH patients. In the hospital model, the hospital delivers all care. In the GP model, the GP is subcontracted and paid by the hospital to manage the patient’s care and utilizes the services of the hospital (nursing care, allied health, pathology, and radiology) to assist with care delivery. In the mixed model, the hospital and GP allocate and share medical responsibility based on particular circumstances such as degree of medical intervention required.
Analysis of cost data
Simple and multi-linear regression analysis was performed using SPSS [27] to estimate univariate and adjusted cost ratios of various types of care. The outcome variable was logarithm of cost because of the skewed distribution of cost. Stepwise regression was carried out to identify models that best described the variance in the cost. The entry criterion for the variable was P of F = 0.05, the removal criterion was P of F = 0.10. The goodness of fit of the model was evaluated using the adjusted coefficient of determination (R2 ) because it more closely reflected the variance in the cost. As several independent predictors were used to model cost, the co-linearity between the variables was evaluated using the tolerance values.
Analysis of mortality and LOS data
Mortality rate, calculated as the percentage of patients who had death recorded in their medical record, was compared for HITH and IHC episodes. Potential sociodemographic and clinical predictors of death and LOS were adjusted for using multiple logistic regression modelling in EGRET [28]. Predictor variables analysed are listed below; all variables were modelled, but the final model included only variables that were significant or which are known to be associated with mortality or LOS.
Dichotomous variables were defined for the presence or absence of various conditions, including cardiovascular disease, respiratory disease, diabetes, and malignancies. Dichotomous variables were also created for the presence of E-codes, which are supplementary morbidity codes that describe a cause of injury. ‘Injury’, as defined in ICD-9/10-CM (International Classification of Diseases), is physical injury (fractures and soft tissue injuries), as well as poisonings and toxic effects.
Variables were also created for death (mortality within 1.5 years of hospital episode), re-admission within 28 days to the same hospital, antibiotic treatment, anti-coagulation, chemotherapy, wound management, scans or other imaging performed during the episode, blood tests performed during the episode, and intensive care (ICU) or high-dependency unit (HDU) care.
We defined the variables as folllows: ‘high number of diagnoses’ for episodes where the total number of diagnoses was >75th percentile; ‘high number of procedures’, where the total number of procedures was >75th percentile; ‘long LOS’, where LOS was >7 days (= 75th percentile); and ‘older age’ if age was >62 years (= 75th percentile).
HITH (all) was defined as any HITH care delivered during episode (includes mixed and pure HITH); pure-HITH as the entire admission consisting of HITH care; and mixed-HITH as part of the episode being in hospital and part at home.
Variables were also created for ‘long HITH LOS’, where the LOS in HITH was >5 days (= 75th percentile); ‘nursing subcontracted’, where some or all nursing services in HITH subcontracted to private nursing providers; ‘cost method’, where cost methods other than Transition or Trendstar (cost method not based on bed days); and ‘rural’, where the hospital was rural.
The following models of HITH care were defined: ‘GP model of HITH’, where the GP is subcontracted to manage the patient’s care and utilizes the services of the hospital (nursing care, allied health, pathology, and radiology) to assist with care delivery; and ‘combined model of HITH’, where the hospital and GP share medical responsibility based on particular circumstances such as degree of medical intervention required. In the hospital model, the hospital delivers all care.
Results
There were 31 hospitals represented in the sample. The hospital model of service delivery was used in 17 (49%) programs and 508 (52%) of matched episodes; the GP model was used in five (14%) programs and 52 (5%) of matched episodes; and the mixed model was used in 13 (37%) programs and 417 (43%) of matched episodes. The mean age of patients was 48.9 years (range 0–99 years) for pure-HITH, 46.7 years (range 0–99 years) for mixed-HITH, and 48.2 years (range 0–99 years) for IHC episodes. The mean number of diagnoses per episode was 3.2 for pure-HITH, 5.0 for mixed-HITH, and 4.3 for IHC. Table 1 shows that the clinical conditions by DRG are distributed differently in pure and mixed HITH.
|
Cost of HITH compared with IHC
Table 2 shows the crude cost comparisons between HITH and ICH episodes. There was no significant difference between HITH and IHC, but pure-HITH was significantly cheaper and mixed-HITH (including the in-hospital component of care) was significantly more costly. After adjustment was made for relevant clinical predictors, the cost of episodes of acute care containing an HITH component overall were significantly less expensive than IHC, by 9% (P = 0.04, adjusted cost ratio 0.91, t = 2.1), and pure-HITH episodes were 38% cheaper than all other episodes (mixed-HITH and IHC) (P < 0.001, adjusted cost ratio 0.62, t = 10.7). The adjusted cost of mixed-HITH episodes was not significantly different from that of IHC.
|
Predictors of total cost (HITH and IHC)
The variables HITH, long LOS, and chemotherapy explained 60% variation in TEC. Table 3 shows that the most important clinical predictors of TEC were long LOS (which predicted higher cost), HITH, and chemotherapy (which predicted lower cost). E-codes (injury codes), stay in an ICU or HDU, and the number of diagnoses and procedures also predicted cost. Cardiovascular disease was also associated with higher cost. Variables that did not significantly explain the variation in TEC included age, death, respiratory disease, diabetes, and rural location.
|
Predictors of cost of HITH episodes
The mean cost of pure-HITH episodes was 22% lower compared with mixed-HITH for total HITH cost (P = 0.004, adjusted cost ratio 0.78, t = 2.9).
TEC for HITH was predicted most strongly by the LOS in HITH. Table 4 shows that types of treatment that predicted higher HITH cost were antibiotic therapy and wound management. Types of diseases that predicted higher cost in HITH were cardiovascular and respiratory diseases. The variables pure-HITH, long LOS (HITH), E-codes, blood tests, antibiotic treatment, wound management, high number of diagnoses, death, cardiovascular disease, and respiratory disease explained 28% variation in total HITH cost. The variables that did not significantly explain the variation in total HITH cost included age, diabetes, re-admission, anti-coagulation, chemotherapy, high number of procedures, and rural location.
|
The cost of HITH varied according to the service delivery model of HITH. The GP model of care was 88% more expensive compared with other models (P < 0.001, adjusted cost ratio 1.88, t = 3.91). However, the GP model sample size was small and belonged to a small number of hospitals. For pure-HITH episodes, the combined model cost 24% less compared with other models (P = 0.02, adjusted cost ratio 0.76, t = 2.2).
Predictors of death and LOS
The percentage of patients who had death documented in the medical records, during or after the admission episode, was 5.2% for IHC and 3.8% for HITH. Although there was a trend towards lower mortality in HITH, this difference was not statistically significant (odds ratio 0.71, 95% confidence interval 0.45–1.12, P = 0.12). For pure-HITH this was 4.0% (21 of 529), and for mixed-HITH it was 3.3% (15 of 448). The strongest predictor of death was malignancies (Table 5), followed by long LOS, respiratory disease, high number of diagnoses, and rural location. Wound management predicted a significantly lower risk of death.
|
The mean LOS of pure-HITH episodes was slightly shorter (3.3 versus 4.0 days) than for IHC matches. In contrast, mixed-HITH episodes had a much longer mean LOS (13.6 days) than their IHC matches (5.6 days); however, less than half of the LOS for mixed-HITH episodes were days spent in HITH. Table 6 shows that pure-HITH was the strongest predictor of shorter LOS. HITH (all of mixed and pure) was the strongest predictor of longer LOS. Other predictors of long LOS were diabetes, E-codes, death, anti-coagulation, wound management, ICU or HDU stay, high number of diagnoses, high number of procedures, and older age.
|
Discussion
This study illustrates the importance of adjusting cost for confounding factors, and might explain why many studies have not found a difference in cost between HITH and IHC. We are not aware of any other studies of HITH cost that have adjusted costs to eliminate the effects of confounding factors.
Cost needs to be adjusted because many factors other than HITH or IHC (such as severity of illness, type of illness, and type of treatment) can influence crude (unadjusted) costs. These factors can be independent of HITH/IHC and may be unevenly distributed between HITH and IHC, even in an RCT, so that a simple comparison of the crude cost of HITH and IHC episodes may give invalid results.
We found that although the mean crude cost of HITH episodes was not significantly different from that of IHC, the adjusted cost of HITH was significantly cheaper than IHC. HITH is cheaper, particularly as a form of total episode substitution. The fact that matched eligible patients were available in the IHC group suggests that there is likely to be potential for greater use of pure-HITH.
Our work is consistent with other studies which have shown that the cost of HITH is less than IHC [9,19]. However, when considering the number of conflicting studies on HITH cost, it must be remembered that HITH is a heterogeneous entity and cannot be consistently defined across studies. Most previous HITH costing studies addressed specific conditions and models of care [4,9,10,15–19]. Our study, however, addressed the entire spectrum of HITH care.
HITH is not a homogeneous entity; at the broadest level, the distinction needs to be made between HITH used as total episode substitution (pure-HITH) and as partial episode substitution (mixed-HITH), as these two groups are distinct. The patient mix, treatments, costs, and LOS in pure-HITH and mixed-HITH were very different, with the pure-HITH group being less complex and less likely to be severely ill. The distinction between mixed and pure-HITH does not appear to have been well addressed by other studies.
Most of the variation in TEC was explained by HITH and chemotherapy (which predicted lower cost), and by long LOS (which increased cost). As expected, increased cost was also associated with the factors that were markers of severity of illness, injury, and clinical complexity. Types of treatment that predicted higher HITH cost were antibiotics and wound management. Types of diseases that predicted higher cost in HITH were cardiovascular and respiratory disease. E-codes (injury codes) also predicted cost, suggesting that injury was associated with higher episode costs. Stay in an ICU or a HDU, and the number of diagnoses and procedures also predicted cost, probably because they are markers of the clinical complexity of cases.
Our results suggest that the cost advantages of HITH are especially pronounced in chemotherapy patients. This finding is different from other studies [18]. Although chemotherapy was a major predictor of lower episode cost, most patients receiving a whole course of chemotherapy require multiple episodes of care, so that on a patient basis, rather than an episode basis, chemotherapy may not be cheaper. This is in contrast to conditions such as cellulitis, which usually require a whole course of antibiotic therapy in one episode of care. On an episode-based comparison, cellulitis would be more expensive than chemotherapy, but on a patient-based comparison those differences may disappear, or even be reversed. There are additional factors other than cost that determine suitability for HITH, including levels of family support, and patient lifestyle and preference. Cancer patients may prefer to receive chemotherapy at work, home, or an oncology day centre.
Age did not predict cost of HITH. This indicates that patient age should not be a barrier when determining suitability for HITH. However, there are factors other than cost that determine suitability for HITH, including patient willingness and levels of family support.
We found that the combined model of care proved to be significantly less expensive (24% cheaper) than the alternatives, and the GP model was 88% more expensive. Most of the pure GP models of HITH were in rural locations, but when this was adjusted for, the GP model remained a significant predictor of cost. This indicates that the GP model, as practiced in Victoria, may not be efficient.
LOS in pure-HITH was slightly shorter than in matched IHC. In contrast, LOS was much longer in mixed-HITH than in matched IHC episodes, because many patients in mixed-HITH had complex conditions and were being transferred to HITH after a longer length of in-hospital ward stay. In addition, the mean LOS of the in-hospital ward component in mixed-HITH was influenced by a high proportion of ‘outlier’ episodes. The other predictors of LOS found in this study are consistent with the literature. Previous studies indicate that LOS may be affected by independent patient factors, such as age and co-morbid conditions) [29–31], procedure-related complications [32], and independent hospital factors (such as staffing and bed numbers). In addition, difficulty of clinical management and severity of illness have been shown to be correlated with LOS for general and specific in-patient populations [33,34].
This study was not designed to evaluate the appropriateness of LOS for specific conditions. However, there is no evidence in this study that HITH care resulted in a longer LOS. This finding is in contrast to other studies, which did not adjust for confounding factors. A knowledge of which conditions and treatments predict longer long LOS is useful in planning an efficient HITH program. Most of the predictors of long LOS are probably markers of increased clinical complexity of cases, and are consistent with the literature on predictors of long hospital stay.
There was no difference in mortality during the episode of care between HITH and IHC patients. This suggests that HITH is safe for appropriately selected patients. Our findings are consistent with several studies, including some that used a RCT design, that have failed to detect a difference in patient outcome for patients using HITH and IHC [7,10]. We did describe some conditions that predicted a higher risk of death, including rural location. Studies have described higher rates of mortality and adverse events in rural hospitals compared with metropolitan hospitals [35,36], and there is some evidence that patients in rural hospitals are less likely to receive state of the art care [37]. However, there may be confounding factors that we have not accounted for which explain the higher mortality in rural hospitals.
A limitation of our study is that we did not measure mortality that occurred outside of the episode of care. The question of long-term mortality is best addressed by a prospective RCT design. However, this is difficult to do with a heterogeneous patient group, and has been achieved mostly with homogeneous groups, where a single treatment type or a few treatment types can be compared in HITH and IHC [10]. Such studies also require rigid inclusion criteria (such as all patients having the same condition and being treated with the same therapy) to make the RCT methodology feasible; such criteria would necessarily exclude a large proportion of HITH patients, and so RCT methodology cannot adequately represent the entire spectrum of HITH.
A further limitation of our study is that the matching process resulted in only 977 of the original 1600 episodes of HITH (61%) being adequately matched to IHC episodes. The study is therefore not representative of the unmatched episodes of care, which may well be different from the matched episodes. However, our study does provide a more generalized view of HITH than most previous studies, as our only exclusion criterion was if the HITH episode could not be matched to an IHC episode for comparison. The retrospective collection of data is also limited by the quality of medical record data. However, we used used multiple data sources, including medical records, hospital morbidity data, and clinical costing data, so that some validation of data was possible.
Finally, our study does not address the cost to families, carers, and the community when using HITH care. HITH does involve cost-shifting from the hospital to the community, but a separate study is required to enumerate that cost.
In conclusion, our study illustrates the complexity of HITH, and the need to account for the numerous confounding factors that can impact on cost. It shows that HITH does result in cost savings compared with IHC, and delineates conditions and treatments that are particularly suitable for HITH. A rational, well planned HITH program can be an excellent means of providing selected types of acute care.
It is likely that HITH will become more cost-efficient and will gain wider acceptance as a mode of delivery of acute care. In addition, more sophisticated models of HITH may develop with increasing experience of HITH programs, with different models evolving for HITH as total episode substitution and partial episode substitution. In addition, future investment in new, less invasive technologies that reduce the need for management in-hospital will increase the range of conditions suitable for treatment with HITH care.
Acknowledgements
The authors wish to thank Alison Yum, Donald Campbell, Jason Wasiak, Craig Williams, and Eugene Chandraraj for their assistance. We thank the Advisory Committee, and people who assisted at hospital visits and medical records departments. The study was funded by the Department of Human Services, Victoria.
Address reprint requests to C. Raina MacIntyre, National Centre for Immunisation Research & Surveillance of Vaccine Preventable Diseases, Children’s Hospital at Westmead, Westmead, NSW 2145, Australia. E-mail: RainaM{at}chw.edu.au 
Received for publication . Accepted for publication March 29, 2002.
REFERENCES
- Iliffe S. Hospital at home: a substitution technology that nobody wants? Br J Health Care Manage 1995; 316: 1761–1762.
- Marks L. Home and Hospital Care: Redrawing the Boundaries. London: King’s Fund Institute, 1991.
- Montalto M, Dunt D. Delivery of traditional hospital services to patients at home. Med J Aust 1993; 159: 263–269.[Web of Science][Medline]
- Grayson L, Silvers J, Turnridge J. Home intravenous antibiotic therapy: a safe and effective alternative to inpatient care. Med J Aust 1995; 162: 249–252.[Web of Science][Medline]
- KPMG Australia. Hospital in the Home Evaluation. Melbourne: Department of Human Services, 1996.
- KPMG Australia. Hospital In The Home 1997/98 Service Audit Final Report. Melbourne: Department of Human Services, 1999.
- Donald IP, Baldwin RN, Bannerjee M. Gloucester hospital-at-home: a randomized controlled trial. Age Ageing 1995; 24: 434–439.
[Abstract/Free Full Text] - Cummings JE, Hughes SL, Weaver FM et al. Cost-effectiveness of Veterans Administration hospital-based home care. A randomized clinical trial. Arch Intern Med 1990; 150: 1274–1280.
[Abstract/Free Full Text] - Hughes SL, Cummings J, Weaver F, Manheim L, Braun B, Conrad K. A randomized trial of the cost effectiveness of VA hospital-based home care for the terminally ill. Health Serv Res 1992; 26: 801–817.[Web of Science][Medline]
- Hill JD, Hampton JR, Mitchell JR. A randomised trial of home-versus-hospital management for patients with suspected myocardial infarction. Lancet 1978; 1: 837–841.[Web of Science][Medline]
- Caplan G, Ward J, Brennan N, Coconis J, Board N, Brown A. Hospital in the home: a randomised controlled trial. Med J Aust 1999; 170: 156–160.[Web of Science][Medline]
- Jennings P. Learning through experience: an evaluation of Hospital at Home. J Adv Nurs 1994; 19: 905–911.[Web of Science][Medline]
- Montalto M, Watts J. Considering the cost of hospital in the home care. Melbourne: Department of Human Services, 1998.
- Fulop NJ, Hood S, Parsons S. Does the National Health Service want hospital-at-home? J R Soc Med 1997; 90: 212–215.[Abstract]
- Dranove D. An empirical study of a hospital-based home care program. Inquiry 1985; 22: 59–66.[Web of Science][Medline]
- Hensher M, Fulop N, Hood S, Ujah S. Does hospital-at-home make economic sense? Early discharge versus standard care for orthopaedic patients. J R Soc Med 1996; 89: 548–551.[Abstract]
- Coast J, Richards S, Peters J, Gunnell J. Hospital at home or acute hospital care? A cost minimisation analysis. Br Med J 1998; 316: 1802–1806.
[Abstract/Free Full Text] - Lowenthal R, Piaszczyk A, Arthur G, O’Malley S. Home chemotherapy for cancer patients: cost analysis and safety. Med J Aust 1996; 165: 184–187.[Web of Science][Medline]
- Beck Friis B, Norberg H, Strang P. Cost analysis and ethical aspects of hospital-based home-care for terminal cancer patients. Scand J Prim Health Care 1991; 9: 259–264.[Medline]
- Board N, Brennab N, Caplan G. A randomized trial of the costs of hospital as compared with hospital in the home for acute medical patients. Aust NZ J Public Health 2000; 24: 305–311.[Web of Science][Medline]
- Jones J, Wilson A, Parker H et al. Economic evaluation of hospital at home versus hospital care: cost minimisation analysis of data from randomised controlled trial. Br Med J 1999; 319: 1547–1550.
[Abstract/Free Full Text] - Shepperd S, Harwood D, Gray A, Vessey M, Morgan P. Randomised controlled trial comparing hospital at home care with inpatient hospital care. II: cost minimisation analysis. Br Med J 1998; 316: 1791–1796.
[Abstract/Free Full Text] - Clarke A. Benefits and drawbacks of hospital-at-home schemes. Prof Nurse 1997; 12: 734–736.[Medline]
- Iliffe S. Hospital-at-home: buyer beware. J R Soc Med 1997; 90: 181–182.[Web of Science][Medline]
- Shepperd S, Iliffe S. Hospital at Home. Br Med J 1996; 312: 923–924.
[Free Full Text] - National Coding Centre. ICD-9-CM. The Official NCC Australian Version. Sydney: Faculty of Health Sciences, University of Sydney, 1995.
- SPSS Inc. SPSS. Chicago, IL: SPSS Inc., 1994.
- Cytel software. EGRET. Seattle, WA: Cytel Software Corporation, 1985–1993.
- Emberton M, Neal DE, Black N, Harrison M, Fordham M, McBrien MP. The National Prostatectomy audit: the clinical management of patients during hospital admission. Br J Urol 1995; 75: 301–316.[Web of Science][Medline]
- Paul SD, Eagle KA, Guidry U et al. Do gender-based differences in presentation and management influence predictors of hospitalisation costs and length of stay after acute myocardial infarction? Am J Cardiol 1995; 76: 1122–1125.[Web of Science][Medline]
- Cleary PD, Greenfield S, Mulley AG et al. Variations in length of stay and outcomes for six medical and surgical conditions in Massachusetts and California. J Am Med Assoc 1991; 266: 73–79.
[Abstract/Free Full Text] - Wolfe MW, Roubin GS, Schweiger M et al. Length of hospital stay and complications after perculateous transluminal coronary angioplasty. Clinical and procedural predictors. Circulation 1995; 92: 311–319.
[Abstract/Free Full Text] - Kelleher C. Relationship of physician ratings of severity of illness and difficulty of clinical management to length of stay. Health Serv Res 1993; 27: 841–855.[Web of Science][Medline]
- Reiley P, Howard E. Predicting hospital length of stay in elderly patients with congestive heart failure. Nurs Econ 1995; 13: 210–216.[Web of Science][Medline]
- Vu HD, Heller RF, Lim LL, D’Este C, O’Connell RL. Mortality after acute myocardial infarction is lower in metropolitan regions than in non-metropolitan regions. J Epidemiol Community Health 2000; 54: 590–595.
[Abstract/Free Full Text] - Ansari MZ, Costello AJ, Jolley DJ, Ackland MJ, Carson N, McDonald IG. Adverse events after prostatectomy in Victorian public hospitals. Aust NZ J Surg 1998; 68: 830–836.[Web of Science][Medline]
- Lim LL, O’Connell RL, Heller RF. Differences in management of heart attack patients between metropolitan and regional hospitals in the Hunter region of Australia. Aust NZ J Public Health 1999; 23: 61–66.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  S. Dumont, P. Jacobs, K. Fassbender, D. Anderson, V. Turcotte, and F. Harel Costs associated with resource utilization during the palliative phase of care: a Canadian perspective Palliative Medicine, December 1, 2009; 23(8): 708 - 717. [Abstract] [PDF]
|
|
|
|
 |
|
 C. D. Armstrong, W. E. Hogg, J. Lemelin, S. Dahrouge, C. Martin, G. S. Viner, and R. Saginur Home-based intermediate care program vs hospitalization: Cost comparison study Can Fam Physician, January 1, 2008; 54(1): 66 - 73. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||