Influence of weekday of admission and level of distress on length of hospital stay in patients with low back pain: a retrospective cohort study

We included all patients with LBP admitted to a medical unit at the Winterthur Cantonal Hospital during a 12-month period (February 1, 2019, to January 31, 2020). Winterthur Cantonal Hospital is a public acute care hospital with over 28,000 inpatient visits a year. In 2019, the medical units had 8216 inpatients [15].

We identified patients with LBP admitted to medical units on the screening list. All patients received pencil and paper questionnaires for assessing health-related personal data at the initial physiotherapy consultation. We asked patients to complete the questionnaire and sign the informed consent document concerning the use of health-related personal data for research purposes. Patients returned the completed questionnaires to their physiotherapist.

The analyses in this study were based on two different samples (see Fig. 1). To analyse the effect of the weekday of admission on LoS (Part 1), we extracted data from the hospital’s electronic medical records using patient identification numbers. To record diagnoses related to specific causes of LBP (‘red flag’ pathologies), we investigated the medical records as well. As specific causes of LBP, we coded cancer, infection, trauma or inflammatory diseases such as spondylarthritis [16]. To analyse the influence of the weekday of admission and patient-reported distress on LoS (Part 2), we only used data from patients who signed the consent document. We did not include patients hospitalized for more than 1 month. In these cases, we assumed that significant complications occurred for reasons that cannot be explained by the initial LBP problem. The further use of routine health-related person data for this study was approved by the regional ethics committee (KEK ZH: 2020–01,465). All analyses were performed in accordance with guidelines and regulations from the regional ethics committee.

Patients with LBP admitted to medical units at Winterthur Cantonal Hospital receive care according to the clinical LBP pathway, with the aim of facilitating a patient-centred interdisciplinary approach (medicine, physiotherapy and nursing). The pathway requires that a specialized physiotherapist assess all patients. Physiotherapy was scheduled on the second day of hospitalization. Immediately after the first physiotherapy session, a consultation between the physiotherapist and the physician in charge should take place. This formal meeting focuses on the diagnosis, further inpatient procedures, and discharge management.

The general aim of the interdisciplinary approach is to achieve a mutual understanding of the pain problem and to set common goals together with the patient. This interdisciplinary procedure cannot be provided over the weekend. On weekends, interdisciplinary meetings are often not performed in general medicine units, as there is a reduction in the ratio of senior staff to patients on weekends [17]. For admissions on Friday afternoon, Saturday and Sunday, the physiotherapy assessment and the interdisciplinary meeting are postponed to Monday. On weekends, specialized physiotherapy assessments are not provided.

The physiotherapy assessment (60 min) is focused on hearing the patient’s full story regarding the cognitive and emotional experiences of their pain problem. Physiotherapists aim to explore patients’ beliefs, emotions, and stress responses associated with their current pain problem as well as their strategies for coping with pain and distress. Furthermore, the assessment addresses fear-related movements or avoidance behaviour. This functional analysis aims to identify maladaptive movements or postures, including extensive protective muscle activation or dysfunctional pain behaviour. In a collaborative process with the patient, physiotherapists explore patients’ ability to relax trunk muscles and to normalize pain-provocative postural and movement behaviours. Physiotherapists and physicians screen for specific causes of LBP (‘red flag’ pathologies).

Physicians are responsible for inpatient management, clinical diagnosis, diagnostic imaging, pharmacological treatment and invasive measures (e.g., surgical procedures). Attending physicians and nurses see patients daily during medical rounds, focusing on diagnosis, therapeutic procedures, and discharge management. Physiotherapists do not routinely take part in these rounds. However, information from the physiotherapy assessment is incorporated into the communication between the patient and health care professionals.

We used the numeric rating scale (NRS, scale: 0–10) to measure average pain intensity over the last week. By means of the German version of the Roland Morris Disability Questionnaire (scale: 0–24), we measured back-specific function [18]. To evaluate patient-reported psychological distress, we used the distress scale of the German Four-Dimensional Symptom Questionnaire (4DSQ). The 4DSQ has been shown to be a valid self-reporting questionnaire to measure distress, depression, anxiety and somatization in patients treated in primary care [19]. The German version of the 4DSQ has previously been validated in a sample of multimorbid elderly people [20]. The questionnaire addresses the presence of symptoms during the last two weeks. Psychological distress was conceptualized as a direct manifestation of the effort people must exert to maintain their psychosocial homeostasis and social functioning when confronted with taxing life stress [19, 21]. In the four-dimensional model of the 4DSQ, distress is conceptualized as the most basic, most general or “normal” expression of psychological problems [19]. Higher scores on 4DSQ scales represent higher symptom severity.

For the analysis in Part 1, we used a negative binomial model for LoS (dependent variable) with weekday of admission as a predictor. We tested the Poisson model against the negative binomial model using likelihood-ratio tests. Planned contrast was used to estimate the difference in log-expected LoS between the weekday of admission dichotomy (Friday/Saturday versus Sunday-Thursday). For Part 2, a negative binomial model was fitted for LoS (dependent variable) with weekday of admission, continuous patient-reported distress (4DSQ distress scale), and their interactions as covariates. We constructed the planned contrast of interest – the difference in (per unit) distress effect between group 1 (admissions Friday/Saturday) and the reference group (admissions Sunday-Thursday). This contrast represents the influence of the distress effect on LoS by day dichotomy (Friday/Saturday versus Sunday-Thursday). The level of statistical significance was set at 0.05. We performed all analyses using R statistical software [22].

During the 1-year period, 177 patients were screened. Four patients with LBP were hospitalized for more than 1 month (extreme cases: range = 31 to 54 days) and were excluded from the study. The mean age of the total study sample (Part 1: N = 173) was 66.10 years (standard deviation, SD = 16.21). Out of 173 patients, 103 were female (59.5%). In 20.8% of patients (N = 36), specific causes for LBP were diagnosed, most frequently fractures (N = 15), followed by cancer (N = 9), infection (N = 6), spondylarthritis (N = 4), and cauda equina (N = 2).

The overall mean LoS of all patients with LBP (N = 173) hospitalized in general medical units was 7.83 days (SD = 5.59). Table 1 provides descriptive statistics on LoS by weekday of admission in patients with LBP.

There was a difference in patients with LBP concerning the mean LoS between admissions on Friday and Saturday compared with other weekdays. The difference in the mean LoS between Saturday and Sunday was 3.5 days.

Figure 2 illustrates the LoS predictions on the log scale by weekday of admission in patients with LBP. Planned contrasts in the coefficients for weekdays showed that being admitted to the hospital with LBP on Friday or Saturday was associated with a significant difference in LoS compared with being admitted between Sunday and Thursday, where log ratio = 0.42 (95% CI = 0.21 to 0.63), corresponding to a rate ratio of 1.52 (95% CI = 1.24 to 1.87) on the response scale.

Figure 3 illustrates the estimated interaction effect (weekday of admission*distress) on LoS based on the negative binomial model in Part 2. The graph revealed that patient-reported distress had no or minimal influence on LoS when patients were admitted on Sunday, Monday, Tuesday, Wednesday, or Thursday. However, patients with LBP with high levels of self-reported distress admitted on Friday or Saturday stayed longer in the hospital than those reporting lower levels of distress. Planned contrast revealed that the difference in (per-unit) distress effect on LoS was significantly influenced by dichotomic day of admission (Friday/Saturday vs. Sunday-Thursday; contrast = 0.816 (95% CI = 0.03 to 1.60) on the response scale.

Although real-world interventions are necessary before we can derive clinical recommendations, the current study provides preliminary evidence for the importance of immediate patient-centred LBP management. Providing multidimensional and interdisciplinary LBP management might be easier in specialized units. However, not all hospitals have specialized inpatient units for patients with musculoskeletal pain. Our findings indicate that even with an interdisciplinary team including medical doctors, physiotherapists, and nurses, LoS can likely be shortened if the interdisciplinary approach is provided within the first days of hospitalization. However, in general medicine units at primary care hospitals, diagnostic and therapeutic services are usually reduced on weekends. Although organizational changes are difficult to implement, strategies to strengthen interdisciplinary LBP management during the weekend should be explored to reduce LoS in patients with LBP. Shortened LoS might offer the potential to reduce direct and indirect healthcare costs associated with LBP.

Some limitations concerning our study design should be considered. First, we used a retrospective rather than a prospective design. To investigate the value of a multidimensional physiotherapy assessment in inpatient care for patients with LBP, a prospective, controlled design is required. Second, the study was conducted in only one specific clinical setting, limiting the external validity of our results. Third, patient data were not measured immediately at the time of patient entry but after the first physiotherapy session. The patient questionnaire used for measuring distress aims to capture patient symptoms over the last two weeks. This should describe the distress level at the time of entry into the hospital; nevertheless, the delayed distribution of patient questionnaires could have distorted measures of psychological factors. Furthermore, the study included patients with nonspecific LBP and with specific causes of LBP. Possible differences between these patient groups were not explored. The heterogeneity of the study sample should be considered. Last, it should be taken into account that the negative binomial model on LoS was based on the data of the 86 patients who completed the 4DSQ. Nevertheless, we consider the estimations derived from this model for LoS with interactions as covariates as robust estimates of effects in the total study sample. Despite these limitations, it can be concluded that the LoS of patients with LBP is significantly prolonged if it is necessary to wait more than 2 days for interdisciplinary and multidimensional LBP management to begin. The influence of prolonged waiting time with a weekend admission might be strongest in patients reporting high levels of psychological distress.

Inpatient care for patients with LBP remains a scarcely researched area. Our results suggest that the level of patient-reported distress influences the inpatient management of patients with LBP. Future research should further explore how patients’ psychological distress relates to therapeutic processes and treatment outcomes. Objective measures of individual suffering could help to systematically capture patients’ cognitive and emotional experiences of pain and to develop strategies for patient-centred LBP management in primary care hospitals. For this purpose, the Pictorial Representation of Illness and Self Measure (PRISM) could be useful when measuring patients’ individual experience of the relationship between their pain problems and their personhood [26‐28]. The tool has been used in patients with rheumatoid arthritis to measure salient aspects of their personhood, which have been affected by the illness, and to estimate how this experience contributes to their suffering [29]. A systematic assessment of patients’ individual suffering would be helpful to evaluate LBP management and to develop patient-centred interventions. Furthermore, future research should investigate why patients with LBP are hospitalized in primary care hospitals. In our study, only approximately 20% had specific LBP with diagnosed serious spinal pathologies such as fractures, infections, cauda equina or cancer. For the other subgroup of patients with nonspecific LBP, it remains unclear why they were hospitalized and not treated in outpatient care settings. More knowledge about their characteristics and needs could help to improve inpatient care and to clarify possible care gaps in outpatient settings for this patient group. Finally, further research is needed to explore the influence of specific diagnoses and multimorbidity on LoS in patients with LBP. This could help identify possible subgroups of patients who need special attention at general medicine units, especially at weekends.