Factors associated with the workload of health professionals in hospital at home: a systematic review

Different health care models have been labelled as hospital at home (HaH) in the international literature over time. Such models vary according to the general organisational guidelines of the health system in each country and according to the funding and the profile of the patients (or diseases) for which care is provided [1‐3]. Heterogeneity is also observed in the type of team in charge of coordinating the care (e.g., hospital vs community-based), the care team composition (e.g., nurses, physiotherapists, occupational therapists), the programme components (e.g., additional elements such as patient and caregiver education), the number of home visits, the availability of after-hours support and specific medical services provided (e.g., home oxygen, intravenous fluids) [4, 5].

In addition, scientific evidence shows a broad terminology to refer to HaH. UK Hospital at Home Society [6] and some authors have defined HaH as “ the care service offered for a limited period of time that provides active treatment by health care professionals in the patient’s home, for a condition that otherwise would require acute hospital in-patient care” [7‐9]. In Spain, HaH is defined as “a care alternative that consists of an organized model capable of providing a set of medical and nursing care and attention of hospital rank (provided by health professionals and material resources of the hospital itself), both in quality and quantity to patients at home, when they no longer require hospital infrastructure but still need active monitoring and complex care [3].

There are two main types of HaH programmes—early supported discharge (ESD) and admission avoidance (AA) [5]. ESD aims to accelerate the discharge of admitted patients, thus, partially substituting hospital care. AA directly admits patients into HaH based on general practitioner referrals, thereby avoiding physical contact with the hospital, or through direct admissions from the emergency room without inpatient stay. According to the systematics reviews conducted by Shepperd et al. [8, 10], there are some benefits that could be obtained with this type of service: reductions at 6 month mortality, rate of hospital readmissions and average hospital stay; improved daily living functional outcomes and quality of life; reduced stress and additional burden on caregivers; and finally reduced costs. Nevertheless, there is a limited evidence on the effectiveness of these types of programs as an alternative to inpatient care and the cost-effectiveness is uncertain. Difficulties relating to definitions of HaH may have reduced the effect attributable to substitution [7, 9], although some benefit was observed [11]. Other authors have reviewed the effect of HaH schemes for patients with specific conditions, such as COPD and heart failure and have found that HaH may be advantageous with respect to readmission-rates in these patients [12‐15].

The availability of resources in the HaH service depends on the number of users, as well as the workload that these users entail. Understanding the factors related to workload, such as the volume and type of care administered to the patients, will offer a better knowledge of the HaH operational mechanisms and will indicate areas for improvement. Thanks to this information, HaH managers could anticipate the needs, as well as support decision making regarding the weight and composition of the teams in the short or long-term. This knowledge could indeed lead to a more appropriate and efficient use of the HaH services.

There are a few studies aimed at identifying the factors that influence workload or workload intensity [16, 17] associated to home-based care and the literature on HaH in particular is scarce. So we conducted a systematic review to identify the factors associated with workload in HaH defined by the Spanish HaH society.

A systematic literature review was conducted following the principles of the PRISMA statement [18].

The study was approved by the local Ethics Committee. A study protocol was prepared, but the review was not registered. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

For the present systematic review, 2,015 references were retrieved: 537 from MEDLINE, 844 from EMBASE, and 634 from CINAHL. After automatically removing duplicated articles, references were reduced to 1,646. The revision of titles and abstracts resulted in 139 potentially relevant references after removing 1,507 records that did not answer the research question and those that did not fulfil inclusion criteria. After full-text assessment 129 studies were excluded (102 did not meet the selection criteria, two were duplicated and no full text was available for the remaining 25 studies). Eight more articles were identified at this stage by inverse search. No papers meeting the selection criteria were located within the grey literature. Therefore, 18 studies were finally included in the present systematic review (Fig. 1).

The reviewed literature shows that, according the used definition of HaH, highly complex patients with time-intensive needs require a significant contribution of health professionals with higher intensity [45, 46]. However, in addition to these patient-oriented measures, the amount of work required for each case should also be considered. Both, the workload and its predictive factors were evaluated in different ways across the included studies. This is probably due to the type of patients, organisations and health professionals studied. In the present systematic review, workload was mainly studied in nurses but not in physicians. The reviewed studies were focussed in home health services and included a very diverse range of patients.

According to the results derived from the included studies, male sex, greater age, worse functional status, greater number of diagnoses and living a rural environment are some of the independent predictors of workload.

Similar results were also found in other studies. Thus, according to a longitudinal study carried out in 1,068 patients over 64 years from a home care program [47], risk factors for receiving more nursing visits at home were male gender, dependency for daily activities decubitus ulcers and receiving emergency medical care at home. In contrast, patients with major cognitive impairment had a lower probability of receiving nursing visits at home. Marek et al. [48] showed that age < 65 years was significantly correlated with the hours of nursing care. Medical conditions such as, mental disorders, musculoskeletal diseases and endocrine disorders, as well as sociodemographic factors, were related to visits, hours or days in home health care [48, 49].

The patient classification systems used to predict workload were valid and reliable, but they had some limitations for their generalisation to different settings. Tools such as CHIRS [31, 36] (medium quality) or ONI [43] (medium quality) showed agency or region-specific differences that might limit the generalisation and the applicability of the data.

Therefore, several factors related to disease severity and comorbidity have been identified as being associated with workload, although these results are very much influenced by the organisation system used and by the assessment of the patients’ needs carried out [48‐50].

In the present systematic review, the studies by Trisolini [44] and Ballard [41] showed that the characteristics of the agencies were a significant predictive factor of the workload, although without clearly identifying those specific characteristics and what role they play. The authors suggest that the effect of the agencies´ characteristics could be due to administrative procedures, the training and practice style of the health professionals, the method of payment (per visit or per hour) and other organisational factors that should be explored and analysed in greater detail. The results of the study by Marek et al. [48] indicated that private insurance, as primary payment source, was a significant predictor of home health care visits, accounting for 3% of the variance.

Among the 18 included studies only two strictly corresponded to HaH definition used by Shepperd et al. [7]. This was because with the heterogeneity of HaH models and the terms we found in the available scientific evidence, we established a broad definition of HaH based on the definition by Spanish HaH society. Nevertheless, despite the great diversity of care models, some common characteristics are shared by all most of them. Therefore, the identified factors related to workload of professionals would be useful and applicable to many care models although not to all.

We also found different ways of measuring `workload´, which complicated the assessment of the comparability of the studies and decisions on staffing needs and costs [27, 44]. First of all, the type of activity performed by health care workers (doctors and nurses) in a HaH unit may be different depending on the composition of the care teams, which undoubtedly impacted on the associated workload. We did not found studies that measured the workload of medical doctors in HaH. Secondly, it should be borne in mind that there are different ways to measure workload: through the numbers of visits or through time spent with each patient. A limitation of using the number of visits is that it is assumed that each visit lasts the same amount of time and thus, does not account for the variation in visit time. In addition, time can be measured as direct or indirect time. Moreover, indirect time has a great impact on the total care time as it can represent between 31 and 60% of total care time [42]. We believe that, the frequency of visits and their duration should be measured in order to determine the workload, as well as the direct and indirect time of the health professionals that perform this type of care. However, it is understandable that measuring time is hardly feasible in practice.

On the other hand, work intensity and workload [51, 52] are closely related concepts that are often used interchangeably although they are not equivalent, leading to confusion: the former is a concept closely related to `patient dependence´, `acuity´ and `severity´, while the latter also includes indirect time, factors unrelated to patients such as, contextual and/or organisational factors and staff-related factors [53].

It should be pointed out that due to the diversity of predictors and ways of measuring workload it has not been possible to summarise the results of the included studies numerically. Additionally, a meta-analysis was not conducted since data synthesis was not indicated. Nevertheless, the significant predictive factors for workload at HaH identified in the included studies were summarised in Table 2.

Among the strengths of our systematic review, we could point out that the work was conducted following the standardised PRISMA methodology. It is also important to highlight that studies written in a range or languages other than English were included and the time period established for the search was very broad.

Lastly, most of the 18 included studies used an appropriate design for the assessment of workload predictors (cohort). Twelve of them were moderate quality and the three high quality studies were carried out in recent years. The quality assessment of the included studies was carried out through the application of tools (CASP) that allowed the discussion and consensus between researchers. Slight adaptations were made for the assessment of cross-sectional studies. The CASP tool has been increasingly used in last years for the quality assessment of articles for systematic reviews and meta-analyses, due to its versatility and completeness [19‐24].

Three of the six prognosis factors assessed showed high or moderate certainty evidence, nevertheless all the scales studied as factors showed low evidence.