Study design
For these cross-sectional studies we surveyed GPs in two German federal states and NH staff from all over Germany. Both studies were part of the HOMERN project (HOspitalisations and eMERgency department visits of Nursing home residents), funded by the Innovation Fund coordinated by the Innovation Committee of the Federal Joint Committee (G-BA) in Germany. The project explores, in depth, health care of NH residents with a focus on hospital transfers.
For the GP survey the sample size calculation was based on a UK survey among multidisciplinary healthcare professionals (including physicians and nurses) with direct experience in acute care of NH residents. The respondents considered 55% of hospital admissions inappropriate [39]. For estimating a 95% confidence interval (CI) with a precision of ± 5% (50-60%) (calculation performed with OpenEpi Version 3.01) we needed a sample size of 381 GPs. Assuming a response of 34% as in a previous survey among German GPs [51] a gross sample of 1,121 respondents was necessary. This number was randomly selected from all registered GPs (including general internists working in primary care; approx. n=5,500) listed by the Associations of the Statutory Health Insurance Physicians (“Kassenärztliche Vereinigungen”) in the federal states of Bremen and Lower Saxony. We used the same sample size of originally 1,121 facilities for the survey among NH staff. Basic data of these NHs (name, address) were also randomly drawn from all approx. 11,200 NHs providing long-term care in Germany listed in the Care Navigator provided by the Federal Association of Local Health Insurance Funds (“AOK Pflege-Navigator”). After checking the sample manually for inclusion criteria, we excluded 52 facilities as they were no longer in place or were caring mostly for children, patients in persistent vegetative state or with prolonged mechanical ventilation, resulting in a final sample size of 1,069 NHs.
Both surveys followed an identical methodological approach. We used a number of strategies found to be effective to increase response to postal questionnaires by a Cochrane review [52], including a short questionnaire, follow-up contact, providing a second copy of the questionnaire at follow-up, personalized letters and university origin. The GP data already contained the physicians’ names to which we addressed the questionnaire. Since for the NH survey the letters were preferably addressed to the nursing staff manager, we searched their names manually. If the respective nursing staff manager’s name could not be found, we used the name of the NH director or the executive board instead, if available. Only if no contact person was detectable the questionnaire was addressed to the current nursing staff manager in the respective facility.
In August 2018, we invited the GPs by postal letter with a paper-based questionnaire and sent all of them a reminder letter (with a second copy of the questionnaire attached) after two weeks. The same approach was used for the NHs in January 2019. Data in both surveys were collected anonymously.
Content of the questionnaire
The four-page questionnaires on medical care in NHs, hospital transfers (including ED visits and hospital admissions), and end-of-life care of NH residents was developed by a multidisciplinary research team of health scientists and GPs. It was pretested with non-involved GPs, whose comments were incorporated into the final version. The current article covers the issues regarding hospital transfers for which the same questions were used for GPs and NH staff. This original version of the questionnaire on GPs can be found in the additional file 1, the original questionnaire in NH staff can be found in Strautmann et al. [53].
First, we asked the participants to estimate the proportion of inappropriate hospital admissions and ED visits among NH residents with the question “Taken as a whole, what is the proportion of inpatient hospital stays and outpatient emergency department visits of NH residents you estimate as inappropriate?” (see additional file 1, question no. 4). Second, we framed four statements containing current courses of action and potential difficulties concerning hospital transfers (see additional file 1, question no. 5): (1) “Residents often do not benefit from inpatient hospital stays”; (2) “Nursing staff calls too often the emergency medical service without prior medical consultation”; (3) “After falls of NH residents there is often no alternative than a transfer to hospital”; (4) “Hospital transfer decisions should be taken more cautiously for residents with advanced dementia”. The healthcare professionals should assess these on a 5-point Likert scale ranging from ‘0 = totally disagree’ to ‘4 = totally agree’. The third part dealt with possible areas for reducing the number of hospital transfers which the GPs and NH staff should again rate using a 5-point Likert scale ranging from ‘0 = no relevance’ to ‘4 = high relevance’ (see additional file 1, question no. 6). Drawing from the existing literature [27, 33, 34, 46, 54] and insights from interviews with nurses and GPs in the scope of the HOMERN project we listed the following eight measures: (1) better communication between nursing staff, (2) better communication between nursing staff and GP, (3) better GP’s care/availability, (4) better medical specialist’s care/availability (5) better availability of (medical) resources in the NH (e.g., catheters, rapid diagnostic tests, drugs), (6) more nursing staff, (7) qualification activities for nursing staff, and (8) explicit advance directives (ADs). Besides, the respondents were given the opportunity stating a measure not mentioned before (free-text).
Moreover, the GPs and the NH staff were asked for the following characteristics (see additional file 1, questions no. 11 and 12): age, sex, location of the medical practice or the NH, respectively (≤2,000, ≤5,000, ≤20,000, ≤50,000, ≤100,000, more than 100,000 inhabitants), and number of years working as a GP or in the current position in the NH (nursing management, facility administration, executive board, other), respectively. Furthermore, the GPs were requested for number of residents they care for and the NH staff should additionally report the number of beds in the facility and the distance to the nearest hospital with ED.
Statistical analyses
Exploratory analyses were conducted to compare responses between GPs and NH staff. We used descriptive statistics and calculated frequencies for categorical data presenting as n (%). For continuous data we stated the mean with standard deviation (SD) and the range. The assessed proportions of inappropriate hospital transfers were compared between GPs and NH staff by Mann-Whitney U test. Responses regarding statements containing current courses of action and potential deficits concerning hospital transfers as well as the assessment of possible areas for improvement were compared between both groups using chi-square tests (χ²-Test). We combined the items ‘totally disagree’ and ‘disagree’ as well as ‘totally agree’ and ‘agree’ to one item, respectively. The same applies to the items ‘no relevance’ and ‘minor relevance’ as well as ‘major relevance’ and ‘high relevance’. Since not all respondents answered every question in the questionnaire the analyses were restricted to subjects with no missing values given in the respective questions (presented as n in Table 1 + 2). All statistics were calculated using the SAS programme for Windows, version 9.4 (SAS Institute Inc., Cary, North Carolina, United States).
Since data in both surveys were collected anonymously consent to participate was not required. For both cross-sectional studies, we received waivers from the medical ethics committee of the Carl von Ossietzky University of Oldenburg in Germany (2018-080 and 2018-147).