Non-specific complaints at emergency department presentation result in unclear diagnoses and lengthened hospitalization: a prospective observational study

Up to 20% of patients admitted to any emergency department (ED) present with non-specific chief complaints (NSC), such as generalized weakness, gait disturbance or tiredness [1]. The analysis of a patient’s symptoms is, however, one important cue in diagnosing the patient [1, 2]. Thus, patients presenting with NSC are at an increased risk of misdiagnosis [3]. Indeed, while one out of ten diagnoses is suspected to be incorrect in the general ED population [4, 5], the rate of incorrect ED diagnoses in patients with NSC was found to be as high as 50% [3]. An acute and distinct medical problem requiring emergency medical treatment can ultimately be identified in as much as 51–59% of these patients [6‐8], despite the initially unspecific presentation, that may confuse, mislead or even discourage the treating ED team. Consequently, patients with NSC are at an increased risk of non-favorable outcomes [3, 9, 10] such as an increased likelihood of hospital admission [11] and an increased 30-days mortality [1]. Similarly, Djärva and colleagues also describe a four-fold risk of in-hospital mortality for patients with NSC in their retrospective cohort study from Sweden [11]. Most studies on the issue, however, demonstrate that NSC are much more prevalent in elderly patients [8, 11, 12] which makes it difficult to tell whether the unfavorable outcomes of patients presenting with NSC to an ED truly result from their challenging non-specific chief complaint or are simply a consequence of the patients’ older age. Because elderly patients represent the fastest growing emergency department population [8, 12], NSC are consequently of increasing importance to emergency care.

In summary, patients with NSC are not only personally at higher risk of misdiagnosis, hospital admission or even death, but also present a relevant burden to any medical system. Yet, little is known about differences in presentation, diagnostic workup and outcome [13] as prospective comparisons of the outcome of ED patients with NSC versus specific complaints (SC) are lacking. We therefore address the following research questions in a prospective observational study:

The study protocol for this study has been previously published [14].

During the study period, 14′187 patients presented to the ED of which 894 patients were admitted to an IM ward. Of those admitted, 66 patients were excluded from further analysis because of palliative or social care as main reason for hospital admission, 73 patients because of a surgical main problem and 44 patients because of a secondary transfer with an established diagnosis at ED admission (Fig. 1). Out of the 711 patients finally included in this study, 165 (23.21%) presented with NSC and 546 with SC.

Raters achieved perfect agreement in the classification of chief complaints as either specific or non-specific (kappa = 0.9) and in the assignment of ICD-10-CM codes (kappa = 0.96).

Overall, Patients were on average 65.62 years old, 322 (45.3%) were female, 400 (56.3%) had previous medical contacts and 78 (11%) were seen in a resuscitation bay. A comparison of patients presenting with NSC versus those with SC is presented in Table 1.

Regarding ED outcome, patients with NSC did not stay longer in the ED compared to those with SC (NSC Median = 6.27 (IQR = 3.11) versus SC 6.09 (3.26), p = 0.497). The rate of vague diagnosis at ED discharge was significantly higher in the patient group with NSC at ED admission (30.5% vs. 23.1%, p = 0.039). At hospital discharge, the difference in the quality of their diagnosis was no longer significant between patients presenting with NSC vs. those with SC (9.1% vs. 9.3%, p = 0.531).

Patients with NSC stayed significantly longer in hospital (Median = 6.51 (IQR = 5.85) versus 5.22 (5.83) days, p = 0.005) and had a higher mortality compared to patients in the specific complaint group (12 (7.3%) vs. 20 (7.3%), p = 0.045). This effect is however observable in one-sided testing according to the prior hypothesis.

Of the variables assumed to be predictive of complaint specificity, only a previous medical contact was significantly associated with complaint specificity in univariate logistic regression (Table 2). Of the variables assumed to be a consequence of complaint specificity, length of hospital stay, quality of the ED diagnosis and mortality were associated on a significance level of p < 0.1 with complaint specificity in univariate regression (Table 2). The association with complaint specificity held for all variables except for mortality in multivariate regression (Table 2) on a significance level of p < 0.05.

Non-specific chief complaints are frequent among patients presenting to any emergency department [1]. In this prospective observational study of emergency patients hospitalized to internal medicine wards at a single center, 23.3% presented with non-specific complains. These patients present a diagnostic challenge compared to those with specific complaints [3, 9], indicated by their significantly higher likelihood of previous medical contacts prior to emergency presentation and the low quality of diagnoses received at the emergency room. This finding corresponds well to the results of previous retrospective studies. For example, Peng and colleagues found that in more than half of the patients presenting to their emergency room, the EDs diagnosis was overruled within the next 90 days [4], indicating ED diagnoses of low quality.

Unspecific complaints at presentation may confuse, mislead or even discourage the treating ED team [6‐8], and no guidelines for the diagnostic workup of patients presenting with NSC are currently available [7]. Indeed, diagnostic uncertainty has previously been termed the “hallmark of emergency medicine, [given that] emergency physicians are challenged daily by the vast spectrum and acuity of clinical presentations they diagnose in a data-poor, rapidly evolving, decision-dense environment” [22]. Consequently, particular attention should be paid to patients presenting with (seemingly non-urgent) non-specific complaints.

One consequence of an emergency presentation with NSC is their higher likelihood to receive a descriptive or otherwise low-quality diagnosis in the emergency room. This has to be seen in the context of the fact, that an acute and distinct medical problem requiring emergency medical treatment can ultimately be identified in as much as 51–59% of NSC patients [6‐8]. The need to further develop and improve the low-quality diagnosis of the ED is evident in our study by the significantly longer time patients with non-specific complaints are hospitalized (1 day on average), although the effect is small. At hospital discharge, patients with non-specific complaints do not differ anymore from those with specific complaints with regard to the quality of their diagnosis. A previous Danish study similarly identified frail patients, a group that often presents with non-specific complaints, to be at an increased risk for admission, length of stay and even mortality after visits to the ED [23]. Given the size of the affected patient population and the financial implications of lengthened hospitalization, this finding further emphasizes the need for an increased attention to patients presenting with non-specific complaints. A systematic review of scoring systems recently compared ten different strategies for risk stratification at ED admission and also emphasized the need for more research, as none of the systems predicted all important outcomes with high precision [24]. Possible approaches include screening for “immobility as a vital sign” [25], reassessing patients with low-quality and descriptive diagnoses [23] or the algorithmic implementation of predefined diagnostic packages triggered by chief complaints [26].

Interestingly, patients presenting with non-specific complaints in our study were neither older nor triaged more urgently at ED admission. This lack of differences at presentation between the groups of patients with NSC and SC stands in contrast to previous studies that report an increased number of female and older patients to present with NSC [2, 9, 13, 27]. This discrepancy may result from the fact that we, by choice, only looked at a rather homogeneous population of patients hospitalized to an internal medicine ward. We deliberately limited our investigation to hospitalized ED patients for three reasons: First, we assumed adverse outcomes to be more frequent in hospitalized patients as compared of those discharged from the ED, thus increasing the statistical power of the study. Second, we suspected the age-distribution of hospitalized patients presenting with SC to be comparable to hospitalized patients with NSC, because the risk for hospitalization generally increases with age, regardless of the presenting complaint. Because NSC are more prevalent in an elderly population, we aimed to establish a similarly old comparison group by including only hospitalized patients. Third, the burden on the health care system is larger in hospitalized than in discharged patients, thus increasing the relevance of any potential finding of this study.

A last finding in our study is the observation that although patients with non-specific complaints have a substantially higher in-hospital mortality than those without (7.5% vs. 3.7%), in-hospital mortality is not a consequence significantly associated with complaint specificity in multivariate regression.

Christensen et al. report an in-hospital mortality of 2.38% and a 30-days mortality of 4.32% in NSC patients [23]. Other mortality rates of NSC patients described in previous studies range from a 30-days mortality in Swiss ED patients with NSC of 6% [1] to an in-hospital mortality of 13% in a Swedish ED population [11]. Compared to this mortality rates, the 7.3% in-hospital mortality of NSC patients found in our study population is rather at the lower limit of previously described mortality rates.

In conclusion, our study highlights the importance of the ED patients presenting with NSC, as they have an increased risk of unfavorable outcome. One practical implication of the results described above is that each patient presenting with NSC should trigger an awareness in every ED physician, that an unfavorable outcomes is likely. In addition, the development of a specific treatment algorithm for patients with NSC in analogy to existing recommendations for e.g. chest pain or sepsis may be another possible way to improve the outcome in these patients. On the systems level, an adaption of existing triage scales in order to increase attention to patients with NSC may be a fruitful approach for further investigation [13].