Emergency department visits due to severe falls: comparing patient self- reports and general practitioner records: A cross-sectional study | BMC Geriatrics

In the analysed population of older individuals presenting to the ED after a severe fall, the agreement between self-reported and GP-reported diagnoses was generally poor, as indicated by Cohen’s Kappa values. However, agreement varied by diagnostic group with PD and diabetes mellitus showing relatively high concordance, while conditions like arthrosis of the lower extremities and gait disorders had no agreement. Several factors were associated with a higher risk of disagreement across diagnoses, with concerns about falling being the only factor linked statistically significantly to disagreement across all conditions.

The heterogeneous results regarding agreement between self-reports and GP-reports align with previous studies in different populations of older adults [7, 17]. Higher agreement for conditions such as diabetes mellitus may be due to the significant impact these diseases have on daily life, which likely enhances the accuracy of self-reports [21]. PD often presents with motor symptoms like tremor or bradykinesia, which severely affect activities of daily living and require personalised treatment plans [22]. Similarly, diabetes mellitus is often managed through Disease Management Programmes in Germany, involving regular GP visits [7], which may explain the high agreement in diabetes mellitus reporting observed in our and previous studies [6, 7, 17, 21]. Cancer, known for its substantial impact on individuals but also their families [23], is also reported with high accuracy. Therefore, these conditions can typically be reliably self-reported, even after a severe fall, potentially reducing the need for GP verification in these cases. Nevertheless, further research is needed to confirm these findings, specifically across different countries where variations in healthcare systems and aftercare may impact the level of agreement between self-reports and GP-reports.

Moderate agreement between self-reported and GP-reported diagnoses was observed for heart diseases and stroke, consistent with previous studies [7, 17]. For instance, Hansen and colleagues analysed 3,189 participants aged 65–85 years and reported similar levels of agreement [17]. In contrast to our findings of moderate agreement on lung diseases, Hansen and colleagues found good agreement for asthma and COPD [17]. However, mixed results have been reported elsewhere. Steinkirchner and colleagues, in a study of 589 participants aged 70–95 years, found moderate agreement for asthma but poor agreement for COPD/chronic bronchitis when comparing self-reported and GP-reported diagnoses [7]. While we did not differentiate between asthma and COPD as diagnostic groups but analysed lung diseases together, this could be another possibility as to why there was found a moderate agreement for this diagnostic group. One possible reason for a low level of agreement on asthma, as suggested by Halm and colleagues in their analysis of 198 participants, is that patients with asthma often may not consider themselves to have the condition when they are asymptomatic [24]. This reinforces the idea that variations in the severity of symptoms could contribute to limited agreement [7, 24] and highlight the need for specific inquiries about diseases with fluctuating symptoms. Additionally, seeking GP input on such conditions may improve diagnostic accuracy.

Along with this assumption we found fair agreement for osteoporosis, polyneuropathy, gastrointestinal diseases and tremor, while only poor agreement was detected for PAD, depression, pain, ophthalmic diseases, incontinence, cognitive disorders, vertigo, renal diseases, sleep disorders, orthopedic diseases of the upper and lower extremities and arthrosis of the upper extremities. No agreement was detected for arthrosis of the lower extremities and gait disorders. Like Steinkirchner and colleagues, we found poor agreement between self-reported and GP-reported diagnoses for renal diseases and musculoskeletal diseases [7]. Likewise, vertigo and neuropathies showed low agreement in our study, consistent with findings from Hansen and colleagues [17]. Conditions such as neuropathies and vertigo, which often present with broad and nonspecific symptoms, may have diverse etiologies and can be assessed using different diagnostic criteria [25,26,27]. This variability may contribute to greater discordance between self-reports and GP-reports. Mannion et al., who compared self-reported conditions with pharmacy claims, also observed low agreement for incontinence and emotional disorders such as depression, a finding mirrored in our results [6]. Lower agreement for conditions may stem from less clear diagnostic criteria [28]. Additionally, diseases that do not require frequent monitoring or significantly impact daily life tend to show lower agreement between self-reports and medical records [29]. Conditions such as osteoporosis or PAD, which showed fair to poor agreement in our study, often remain asymptomatic for a long time [30, 31]. This may further explain the high level of discordance observed, as these diseases may not significantly affect activities of daily life, leading individuals to overlook them in self-reports. The complexity of symptoms, the lack of need for regular monitoring or treatment, and the asymptomatic nature of certain conditions suggest that patient education and communication by GPs should be intensified to improve self-reporting accuracy. To streamline the collection of pre-existing diagnoses during multifactorial assessments after severe falls, providing GPs with a predefined list of conditions – such as those mentioned above – could accelerate the process and reduce the burden on healthcare providers.

Higher concerns about falling were a relevant predictor of greater disagreement between self-reported and GP-reported diagnoses. However, no previous studies have specifically examined the possibility of temporary memory loss and the subsequent reduction in self-report accuracy regarding diagnostic information following a fall event. This finding may be particularly relevant to the present study population, as individuals with elevated FES-I scores may focus primarily on the recent fall, potentially neglecting to recall preexisting conditions. One possible explanating is Chung and colleagues’ finding that a notable minority of individuals develop post-traumatic stress disorder (PTSD) following a fall [32], a condition associated with alterations in memory and concentration due to trauma-induced stress [33]. Thus, an elevated FES-I score may be a marker of an increased stress level in some individuals, potentially leading to reduced accuracy in self-reports of preexisting diagnoses.

Another potential factor is mild cognitive impairment, which has been identified as a predictive factor for concerns about falling [34] and may reduce self-report reliability. Prior research has shown low concordance between self-reported and clinically diagnosed dementia in individuals with cognitive impairment [35]. However, in our analysis, MoCA scores were not significantly associated with disconcordance. This may be due to a stronger influence of FES-I or the mitigating effect of proxy assistance in providing medical history, potentially limiting the impact of cognitive status. Giving these findings, involving GPs in multifactorial fall risk assessments may be particularly important in individuals with high FES-I scores. Establishing FES-I cut-off values to identify thresholds at which GP involvement improves diagnostic accuracy could enhance clinical practice. Further research is warranted to validate this approach.

In this analysis, increasing age was identified as a predictive factor for a greater disagreement on PAD, cancer, renal diseases, gait disorders and frailty between self-report and GP-report. Previous studies have shown that older adults are more likely to have discrepancies regarding conditions such as myocardial infarction and stroke [7], and various heart diseases like hypertension and angina [6, 21]. Age-related cognitive decline may explain some of these discrepancies, as the risk of cognitive impairment rises with age [36, 37]. Additionally, the prevalence of multimorbidity increases with age [38], leading to a greater information burden and difficulty in recalling multiple diagnoses. Based on these findings, age should be considered a key factor, along with the short FES-I, in identifying patients who may require diagnostic support from their GP. Although improved data exchange via electronic health records may reduce this need in the future, GPs currently serve as the main point of diagnostic integration across healthcare settings. Incorporating age as one of several factors may help identify individuals for whom GP input would enhance diagnostic accuracy. Future research is needed to establish clear thresholds for when such support is warranted.

The higher risk of disagreement in women diagnosed with osteoporosis, compared to men, aligns with findings from Mannion et al., who reported a 5 times higher likelihood of discordance for osteoporosis [6]. Osteoporosis is often asymptomatic until fractures occur and women are more frequently affected [39]. A study by Lewiecki et al. found that women often underestimate the severity of osteoporosis [39], despite its relevance to fall-related fractures [40]. Therefore, this diagnosis requires particular attention, and a specific question should be included in the risk assessment. Interestingly, higher MoCA scores were associated with disagreement on cancer diagnoses. Although a higher score indicates better cognitive function [10], this discordance suggests that even cognitively intact individuals may provide incomplete diagnostic information. Thus, obtaining details from GPs is crucial not only for those with cognitive impairment.

The predominantly poor agreement between self-reported and GP-reported diagnoses underscores the importance of accurately assessing preexisting conditions as part of the multifactorial assessment following a severe fall. As there is broad consensus that improved communication between physicians and patients enhances treatment adherence [41], health-outcomes [42, 43] and that optimised information delivery improves understanding and recall of diagnoses [44, 45], strengthening doctor-patient communication may also improve the accuracy of self-reported medical histories. Enhanced communication could increase patient awareness of existing conditions, potentially improving interactions in the ED and facilitating early treatment modifications. Additionally, actively involving caregivers or proxies – particularly in high-stress environments like the ED – could provide essential diagnostic insights. However, this approach requires careful considerations of available resources.

Given the challenges in assessing preexisting conditions, digitalisation of healthcare systems is crucial [46], since assessing electronic health information in the ED has been associated with improvements in care processes [47]. In Germany, the electronic health card (eGK) has the potential to enhance clinical decision-making [48] particularly in the ED where timely access to patient history is critical. However, existing barriers, such as the coexistence of paper-based record-keeping systems and various electronic systems in the German healthcare landscape, continue to hinder the widespread implementation of information transfer via the eGK [46]. Further, a comprehensive infrastructure that enables the exchange of clinically relevant data across all healthcare providers must be established [49]. Future research should investigate the eGK as a valuable tool for accessing diagnostic data in ED settings, with proper authorisation protocols in place and full compliance with data protection regulations, including personal authorisation and consent from the insured individual [50].

Limitations and strengths

Unlike other studies that have examined a comparison of concordance of diagnostic information, this analysis is the first to investigate this issue specifically in older adults visiting the ED after a severe fall. The study directly compared participants’ self-reports with the GP-reports, with the GP-report being treated as the gold standard. However, it is important to acknowledge that GPs may not always serve as the definitive source of diagnostic information. GPs may lack diagnostic data collected by various specialists that was not transferred to them, an issue assumed in previous research [7]. In our study GPs reported more diseases apart from ophthalmic diseases which have been detected as the only diagnostic group that were reported more frequently by participants than by their GPs. One possible hypothesis may be that even small losses in visual acuity can be perceived as a major subjective impairment which could justify a higher frequency in the self-reports, but may also indicate a lack of communication between specialists and GPs. Additionally, this analysis included only participants whose GPs provided diagnostic information, potentially introducing a selection bias. GPs who participated in the study by sending diagnostic data may have greater interest in detailed documentation. Diagnoses such as frailty or gait disorders, which are often documented by GPs, may not be perceived by participants as formal diagnoses, but rather as everyday limitations, and therefore may not be reported. Similarly, cancer diagnoses were included regardless of time since diagnosis, remission status, or whether the condition was resolved and no longer impacting daily life. This may partly explain discrepancies between self-reported and GP-reported diagnoses. While such conditions are important due to their implications for fall risk and functional decline [51], clear communication between GPs and patients remains essential to ensure awareness and understanding of these diagnoses. This discrepancy could explain the high disconcordance between self-reports and GP-reports for these conditions. Although diagnoses like frailty are essential to examine, as they increase the risk of future functional decline and falls [51], communication between GPs and patients is particulary important in ensuring patients are informed about such diagnoses. Improved dialogue may help bridge the gap in understanding and recognition of these critical risk factors. Living situation was initially included as an independent variable (categorised as living alone, with others, or institutionalised). However, due to the small number of institutionalised participants, several models produced highly unstable or implausible odds ratios, likely driven by sparse data. As a result, living situation was excluded from the final models for overall diagnoses, cancer, cognitive impairment, sleep disorders, tremor/RLS, and arthrosis of the upper extremities. Future studies should include larger institutionalised subgroups, based on a priori power calculations, to more robustly investigate the potential influence of living situation across a bigger sample.