Based on a national cohort of colorectal cancer patients identified by the Colombian health system, the present study describes the fragmentation of care during the first year after diagnosis and its association with survival for 5,036 patients in the country’s contributory health system. Using the number of different providers who treated each patient for cancer care during the first year, this study describes the distribution of fragmentation for the country, identifies a cut-off point for this distribution and finds that patients exposed to more fragmented networks (with 9 or more different providers) have a 35% higher risk of dying at 3 years. This is the first study of this type in a low- or middle-income country.
The lack of a standard measurement of fragmentation makes it difficult to compare these findings with other studies. Decreased survival at 1 year has been reported for patients with colorectal cancer who had undergone surgery and had been readmitted and seen by a different surgeon than the one who performed the surgery, with an HR of 1.55 (95%CI; 1.40–1.73) (38). In addition, a comparison of academic hospitals showed that 90-day survival for patients with stage II-III rectal cancer is poorer for those receiving fragmented care, with an HR of 1.11 (95%CI; 1.05–1.17) (39). Meanwhile, a study that explored the association between specific survival at 1 year for patients with stage III colon cancer and fragmentation, defined as receiving care in more than one hospital, found no difference in survival, with an HR of 0.98 (95%CI; 0.89–1.07), while the average cost of care was 8% lower (40).
The study herein measures healthcare fragmentation based on a type of fragmentation known as inter-hospital fragmentation, given that it is based on the number of different providers that treated the patient. It used the number of different providers that a patient had to see during the first year after the health system identified them as ill. The literature has reported another pathway to measure inter-hospital fragmentation, which is readmission to a hospital other than the index hospital of care. This has been associated with higher mortality, longer length of stay and a higher risk of readmission (41–43).
The present study found geographical differences in the fragmentation of care. As shown in Fig. 1, the Amazon and Orinoco regions had the least number of cases and low levels of fragmentation, as measured by our study. And no patients were registered in some of the country’s departments. These differences can be explained by the fact that these regions have few specialized cancer care services, and patients can travel to departments with more complex services for diagnosing and treating their disease. In regions with lower population density and fewer services, patients may need to travel long distances to receive care, which may cause delays in diagnosis and treatment. There is a need to develop fragmentation measurements that take into account the distance traveled by patients, especially in countries with marked geographical differences in the supply of healthcare services and access to those services (44).
Healthcare fragmentation can impact the survival of patients with colorectal cancer in multiple ways. It can lead to delays in diagnosing the illness and in initiating treatment, which has been associated with poorer survival (15–17). Patients may receive care from different specialties in an uncoordinated manner, care by multidisciplinary teams has been associated with increased prescriptions for chemotherapy and increased survival for patients with colorectal cancer (18). Specialized cancer care centers have teams that are highly experienced in diagnosing and treating, and for patients with colorectal cancer this has been associated with increased survival (19, 45–48).
The present study used propensity score matching to control for confounding (49). Matching algorithms were selected through an iterative process that made it possible to select the algorithm that produced the best balance of the baseline variables. The sociodemographic characteristics contained in the UPC database and comorbidities were identified with ICD-10 codes to control for confounding for as many variables as possible. Although the algorithm that was selected balanced the observable variables, the balance of unobservable variables was not verifiable.
While our main outcome was overall survival at 3 years, it has also been found that healthcare fragmentation impacts other outcomes, such as hospital stay, readmission and costs, especially for chronic diseases. Studies are needed to explore whether healthcare fragmentation in the case of colorectal cancer impacts economic and clinical outcomes such as length of stay, readmissions, specific mortality and quality of life, among others (7,42,50–52).
The limitations of this study are related to the use of administrative databases, which are not designed for clinical research, and therefore, they do not include clinical variables. We addressed these variables by identifying comorbidities in the Charlson Index, but we did not have information on the staging of the disease upon inclusion in the cohort.