Surgical resection and reconstruction remain the leading treatment modalities for head and neck tumors. However, patients often present with significant comorbidities which can lead to complications during or after lengthy surgical procedures and may require a high-care environment for postoperative recovery. In demand for evidence-based clinical care during the era of limited or constrained financial and human resources in hospitals, it is crucial to implement appropriate and standardized triage for immediate postoperative management of these patients.
The classification developed by Saklad et al. in 1941, modified in 1961, and published by the American Society of Anesthesiologists (ASA), was constructed to preoperatively assess overall health of surgical patients and estimate peri- and postoperative complications [11]. In literature, ASA rating correlates with a higher number of postoperative complications after microvascular reconstruction, such as flap failure or overall survival time [7, 12,13,14,15]. However, Mücke et al. criticized the ASA-score to be sweeping and nonspecific for the variability of head and neck tumor patients [15]. In our study, ASA score as a standardized predictor (OR 0.89, 95%CI 0.68–1.16; p = 0.395) performed worse than the new TRS (OR 1.65, 95%CI 1.32–2.06; p < 0.001) in predicting prolonged IMC/ICU stay (Fig. 6). Preoperative health status is often condensed in various comorbidity scores during preoperative anesthesiologic evaluation. In this regard, the Charlson Comorbidity Index (CCI) uses a weighted cumulative index with categorized comorbidities that showed a significant association with postoperative mortality [16]. In major head and neck surgery overall mortality is rare (0.5–1.5%) and commonly due to myocardial infarction, embolism, or pneumonia [17]. Therefore, focusing on mortality as a possible endpoint parameter, rather than on specific complications such as prolonged IMC/ICU stay, may be one of the reasons why CCI appears to be less relevant in our study population as well (OR 1.15, 95%CI 1.16–1.93; p = 0.338) (Fig. 6). The authors that created the Functional Comorbidity Index (FCI) hypothesized that comorbidities associated with physical function are distinct from those primarily associated with postoperative mortality [10]. Therefore, FCI performs better in predicting physical function than indices predicting postoperative mortality as an outcome [18, 19]. For predicting prolonged IMC/ICU stay, this is both clinically relevant and statistically significant (OR 1.49, 95%CI 1.16–1.93; p = 0.002) (Fig. 6).
In order to account for procedure-specific risk, the ESA/ESC guidelines classify surgical risk by the incidence of cardiovascular-related lethality or the incidence of non-fatal myocardial infarction within 30 days regardless of any patient comorbidities. Incidences < 1% are assessed as low, between 1% and 5% as intermediate, and > 5% as high intervention-specific risk [20]. The Surgical Risk Classification according to John Hopkins, which is also widely used, graduates five groups according to the increasing degree of invasiveness of the surgical procedure. It is interdisciplinary and does not discriminate among the surgical spectrum of different surgical disciplines [21]. However, not only do oral and maxillofacial surgical procedures differ from one another in terms of invasiveness, complexity, blood loss, duration of surgery etc., even tumor surgery in the head and neck region itself offers a wide range of procedures. These can only be stratified vaguely by means of a broad and interdisciplinary score. Especially colleagues from other disciplines sometimes lack differentiation criteria for enhanced resource planning. Ex post, a more precise differentiation can be achieved using the operating time. However, this information is not available preoperatively and cannot be used for predictive risk evaluation and resource allocation. In general, our partners in anesthesia often regard a routine 24-hour monitoring in the intermediate or intensive care unit as necessary and safe [22, 23].
In order to preoperatively calculate the individual probability of a prolonged IMC/ICU stay we combined the two relevant predictors TRS and FCI in order to enable an intervention-specific risk prediction (TRS) in addition to the patient-specific risk constellation for complications (FCI) (Fig. 7). If the TRS and FCI coefficients of the logit function are condensed to the exponential function, the probability PR (IMC/ICU prolonged) can be directly derived [24]:
$$\eqalign}\,\left( }} \right)} & }\,\left( }}\,}\,}\,}\,}}\,}\,}\,}\,}}} \right)\, \cr & }\,\left( }\,\left( }}\,}\,}\,}\,}}\,}\,}\,}\,}}} \right)} \right) \cr}$$
If, for example, a patient with cT2cN0 squamous cell carcinoma of the mandible and anterior floor of the mouth (2 regions = TRS 1) undergoes tumor resection with continuity-preserving mandibular bone resection and reconstruction with radial forearm flap (surgical procedure = TRS + 3), a total TRS of 4 results. With regard to comorbidities, the sample patient presents with osteoporosis and peripheral vascular disease, resulting in an FCI of 2 points. The sample patient thus has a probability of remaining in the IMC/ICU ward for more than 1 day of:
Pr (IMC/ICU prolonged) = exp(0.44 * 4 + 0.29 * 2–2.7) / (1 + exp(0.44 * 4 + 0.29 * 2–2.7)) = 0.411, or 41.1%. With a cutoff value of 0.2, this model of TRS and FCI (ROC area (TRS/FCI) = 0.6701) achieves a better prediction than a prediction model of ASA class and CCI (ROC area (ASA/Charlson) = 0.5965) (p = 0.031) (Fig. 8).
If patients at a high risk of a prolonged IMC-/ICU stay could be identified preoperatively, additional safety measures could be taken or the surgeon may, in some cases, even select an alternative reconstructive strategy. This predictability index could be used in the preoperative setting to improve patient counseling, as well as postoperative organization and treatment algorithms. While most components of risk prediction are not reversible, they do raise awareness among the medical team and can provide patients with realistic expectations regarding their surgery and postoperative course.
Additionally, the risk of postoperative complications does not necessarily equate the need for critical care services. A report from Hong Kong [23] concluded that many anesthesiologists recommend generalized monitoring for 24 h in the ICU after major head and neck tumor surgery often based on defensive medical practice or factors other than medical necessity [25]. In several parts of the world, different surgical specialties perform head and neck tumor surgery either alone or as part of a multidisciplinary surgical team that may include oral and maxillofacial surgeons (OMF), general surgeons, plastic and reconstructive surgeons (PRS), and ear-nose-throat surgeons (ENT). Each of them may contribute a different perspective and expertise to the care and treatment [22]. The British Association of Head and Neck Oncologists asked 253 surgeons from ENT, OMF and PRS about their use of intermediate and intensive care facilities after head and neck tumor surgery [26]. Regarding microvascular flaps, PRS preferred monitoring in an IMC ward, whereas OMF surgeons tended to use the ICU ward. After tracheostomy most ENT and OMF surgeons felt comfortable with care in a regular ward, whereas PRS preferred the IMC. Over all surgical disciplines, one third of the surgeons would start a surgical tumor resection with microvascular reconstruction even if the monitoring ward of their choice (IMC or ICU) would not be available. The survey suggests that there are different approaches and preferences regarding postoperative care after head and neck tumor surgery not only between different surgical departments nationally and internationally but also within the field of maxillofacial surgery itself. Resources, type of hospital care, staff qualifications, availability of IMC/ICU ward facilities, expertise and workload of (nursing) staff and funding issues may be additional factors influencing the postoperative regimen [27]. Thus, it is a potential limitation of our study that we used individual and hospital-specific criteria defining the outcome of “prolonged IMC-/ICU-stay" > 24 hours. The TRS and our risk prediction model may not be directly applicable to other hospitals.
Lack of standardized monitoring protocols or inaccurate assessment of needs can lead to both under- and overutilization of IMC and ICU wards. Monitoring of microvascular flaps can be a trigger for IMC or ICU transfer in some hospitals due to trained nursing staff and more favorable patient-nurse-ratios [28]. According to a survey of plastic surgeons up to 61% of microsurgical centers in the United States prefer to monitor their free flap reconstructions in the ICU. ICU utilization is a major component of hospital resource costs in the United States with significant financial implications. Cornejo et al. [28] found an average length of ICU stay after free flap reconstruction in head and neck surgery of 5.8 ± 0.5 days. Patients with postoperative complications had longer ICU stays compared to patients without complications (8.8 ± 13.4 versus 5.6 ± 15.3 days, p = 0.24). In our study, the difference in the discharge profiles of patients in the individual TRS groups was statistically significant (logrank p = 0.004) (Fig. 4). When the regular IMC/ICU length of stay of 1 day was exceeded, the total hospital length of stay increased by 7.7 days (95%CI 5.35–10.08 days; p < 0.001). In Germany, this fact is already taken into account in revenue calculation. Comorbidities as well as planned and unexpected procedures trigger a complex calculation of a case-specific patient comorbidity complexity level (PCCL), which impacts the case-specific revenue. Likewise, the lower, average and upper hospital length of stay is calculated individually and influenced by PCCL score. Thus, PCCL regulates resource consumption within individual cases and can compensate for additional costs. Comparing the well-established PCCL as a benchmark predictor to our TRS shows a statistically significant correlation (Spearman’s rho 0.24; p < 0.001). Unlike PCCL, our TRS is already available at the time of hospital admission which enables a priori risk stratification and resource scheduling.
In Germany, OMF tumor resection and microvascular reconstruction currently generates a total revenue of EUR 23,583, a lower limit LOS of 6 days, an average LOS of 20.1 days and an upper LOS of 36 days. As we did not want to compare absolute numbers in total LOS, we calculated the proportional LOS considering the actual LOS and case-specific mean LOS of patients with the same spectrum of comorbidities and surgical procedures. We thus created the parameter “long-stay patients” if the proportional LOS was larger than the case-specific upper limit LOS. If a patient exceeds the upper limit LOS the daily revenue is no longer cost-covering. If the regular IMC/ICU LOS of 1 day was exceeded, the probability for also exceeding the upper limit LOS and becoming a “long-stay patient” increased by 70.1% (IRR 1.71; 95%CI 1.02–2.85; p = 0.041). In the German Health Care System exceeding the upper limit LOS is no longer cost-effective, besides the fact that utilization of the ICU is a cost-increasing factor itself. Especially in case of additional postoperative complications the increase in total revenue can remain insufficient while simultaneously more resources are consumed. This jeopardizes cost-effective patient care especially at hospitals providing specialized and maximum care. Already in 2007, Jones et al. [29] urged to directly focus on reducing peri- and postoperative medical complications when striving for cost-effectiveness of microsurgical reconstruction in head and neck tumor surgery. In this context it has to be evaluated whether enhanced care by skilled stuff on general surgical wards can reduce ICU utilization and enable a more targeted allocation of the resource “ICU monitoring” without compromising patient care [22, 30]. According to Haddock et al. [31] the cost per ICU bed in the US amounts to approximately $2,360 per day compared to $900 in a non-ICU setting. In our institution, the cost of ICU monitoring is EUR1,200/day, while a total amount of EUR2,300 for financing the patient’s ICU stay is included in the total case revenue of tumor resection with microvascular reconstruction. Therefore, ICU monitoring is not modeled to cover costs after the 3rd day. In this regard, our index helps both surgeons and anesthesiologist to stratify the risk of a prolonged IMC/ICU stay during the preoperative visit and to adjust resource allocation in specific situations.
Finally, our TRS helps to attribute complication rates and enable a fair comparison of the quality of care in different hospital settings through risk-related adjustments. In 2016, a new law to reform the structures of hospital care came into force in Germany. This mandates the concept of quality-based reimbursement elements in certain specialties. To improve outcome quality additional surcharges and penalty fees for good and inadequate quality of care are being developed. Therefore, specialized and established quality indicators are used. In order to enable a comparative evaluation of the results of different hospitals, the risks of the different patient groups must also be taken into account. It is crucial to compare quality measures regardless of occurring complications on external factors (patient complexity, complexity of surgery), so that an objective risk adjustment can be assumed. This is particularly relevant for hospitals providing maximum care as they cannot perform risk selection [32, 33].
A potential limitation of our study is its retrospective design that can account for observer bias. We are constantly collecting prospective data on patients with head and neck tumor surgery to further validate our index. Additionally, we would like to add that established medical scores such as the ASA classification or the Charlson score have their rightful place in preoperative risk analysis, especially in an inhomogeneous patient collective. Thus, our index certainly cannot be regarded as an exclusive tool, but can be used as a supplementary instrument in the prediction of resource-intensive patient courses, particularly in maximum care hospital settings. Furthermore, the index was created based on the specific circumstances of our hospital and our patient population, whether it can be transferred to other clinics or to countries with other health care systems has yet to be studied.
留言 (0)