Healthcare use by long COVID patients after starting an integrated care unit

A health economic analysis plan was not developed before the study was conducted.

The design of the study consists of a cohort-comparative population. The case group corresponded to all patients who were admitted to the Long COVID unit between June 2020 and December 2022 (n = 341). Patients were diagnosed with the ICD B94.8 code (Sequelae of other specified infectious and parasitic diseases) before the introduction of the U09.9 code for PCC. The only inclusion criteria for the sample under study was for the individual to be over 18 years old, to have a diagnosis of PCC and to reside within the geographic radius of the hospital's service area, specifically, the North Metropolitan Area of Barcelona (Catalonia, Spain). If the patients fitted the inclusion criteria, they would enter the Long COVID unit and their data would be collected.

The control population was obtained through a matching algorithm from administrative records of the reference. The algorithm is based on a criteria of similarity to the case population by the following variables: sex ratio (approximately 2:1 female-male), age-group (an approximate similarity), and socioeconomic status (based on pharmacy copayments). Having acute COVID or its sequelae was not a criteria for the control population selection, in order to maintain the integrity of the observational study.

The resulting cohort had a size of 49,419 patients: 341 from the case population and 49,078 from the matched case population. Both populations' use of services was measured from December 2017 to December 2022.

The study was conducted in the Germans Trias i Pujol Hospital (Spain) and the primary care centres in its area between July 2020 and December 2022. This is the largest monographic Long COVID Unit in Spain. The Long Covid Unit's objective is to diagnose, assess, and rehabilitate individuals suffering from PCC through multidisciplinary care by physicians, nurses, nutritionists and psychologists [11].

The unit's guidelines outline a structured approach to patient assessment and follow-up. The process begins with an initial visit during which the nursing team develops a care plan, offers health education, and evaluates the patient's symptoms. A family doctor then conducts further assessments, such as physical examinations and complementary procedures like chest X-rays, spirometry (SpO2), electrocardiograms, fatigue tests, and functional tests.

Up to six weeks after the initial visit, a first follow-up with the family doctor takes place to assess analytical results and functional test ratings. Based on the severity of the condition and results, patients may be referred to infectious diseases specialists, telematic rehabilitation, or hospitals for further evaluation and treatment. During this period, another visit to the nursing team occurs to provide additional health education and continue the rehabilitation plan.

After six weeks, medical attention is provided either remotely through calls, video calls, or on-site, depending on the patient's needs and the progress of rehabilitation. At 12 weeks (3 months), tracking visits are scheduled to monitor the patient's progress and response to treatment, including clinical analysis results, fatigue and functional tests, and overall condition. Based on these evaluations, patients may continue rehabilitation or be referred to higher levels of healthcare for further treatment. At 24 weeks (6 months), a final visit is conducted to conclude the process.

The costs of PCC patients were compared both in a retrospective manner (pre and post diagnosis and treatment) and in a cohort-comparative setting (in contrast with a comparison population). The retrospective analysis covered a symmetrical 12-month period (six months before and after diagnosis), while the prospective analysis in the control population spanned 12 months.

We also retrospectively compare the use of different services in primary and hospital care, procedures, hospitalizations and emergencies. A cohort-comparative comparison in these services would be difficult to measure due to the different orders of magnitude of the sample sizes and the large amount of missing values in the control population.

This study employs a partial patient perspective to estimate the costs produced by post-COVID19 condition patients. The costs of the patient are paid by the hospital, which is then reimbursed by the public insurer (Catalan Health Service), whose financing comes from taxes paid by citizens.

Regarding the time horizon of the costs, patients were discharged from the Unit after 6 months and subsequently returned to the standard pathway of primary care and specialist visits.

No discount rate was chosen as the time horizon of the costs was under 1 year.

Tables 1 and 2 describe the prices for each form of healthcare resource utilisation. The included resources were primary care visits (in different forms, as specified in the table), outpatient care (first and successive visits of an episode), inpatient care, emergency care and different procedures. Costs were valued per the Catalan Law regulating billable assumptions and concepts and approving the public prices corresponding to the services provided by the Catalan Institute of Health, which acts as a cost catalogue for forms of healthcare resource utilisation [14]. In the results, we present the average number of visits per six months (for the pre-post comparison) and in 12 months (for the cohort comparison).

Due to the staggered entry of patients into the program, a fixed time-frame for studying visits was not feasible. Instead, a balanced approach was adopted, setting an individual threshold of six months before and after diagnosis for each patient. Visits were recorded individually in the databases, then aggregated by service and patient. This method was used to calculate averages by age-group, sex, and the entire population.

In order to monetize different types of services across providers, we valued costs in euros at 2020 prices, following the Catalan healthcare billing law [14]. The reporting of this study follows the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) framework for economic evaluations, as recommended by health economists associations [15].

At baseline, demographic and analytical variables were collected for the case population, while the last available value was used for the control population. The median and IQR are used to describe the quantitative variables, while n(%) are used for qualitative variables. The p-value indicates the statistical significance of the difference between the two groups being compared. To test the significance of the differences, we used Wilcoxon, Fisher, Pearson and Kruskal-Wallis tests. Finally, N/A indicates the missing values for each variable. The software R (version 4.2.2) and its packages tidyverse, glmnet, and zoo were used for the data manipulation and statistical analysis [16 –19].

The recruitments of patients in the Unit was used to better understand their needs and to define the personalized care plans. Clinicians and other healthcare professionals involved in the Unit were part of the planning of the study.

Funding for the study was provided through fundraising campaigns organised by the non-profit foundation Fundació Lluita contra les Infeccions, which included "yomecorono.org" and Gala contra les Infeccions, Editions 2021 and 2022. None of the funding sources played a role in the study's design, data collection, data analysis, interpretation of results, or the writing of the report.

We evaluated 49,078 individuals in the control group and 341 in the case group (Table 3, second column). The Case group was older and had higher values for systolic and diastolic blood pressure, body-mass index, coronary risk, cholesterol, ferritin, and glomerular filtration rate compared to the Control group, with no other significant differences in demographic characteristics. Hypertension and diabetes mellitus type 2 were the most common chronic conditions in the Case population, with several other chronic diseases with shared risk factors and comorbidities (cardiovascular disease, obesity, older age) also being overrepresented.

The third column presents the characteristics of the case population by sex. Sex was included as reported by the physician in the health records. There existed significant differences between male and female users in terms of pharmacy copayment index, systolic pressure, coronary risk, ferritin levels, and glucose levels. The chronic conditions were not broken down by sex given the low N.

We analysed the ratio of visits categorised by age-group and sex (Table 4) from both the retrospective and cohort-comparative perspective. For the retrospective point, the mean number of visits was 2.3 per month before diagnosis and 1.93 after. The reduction was greater on men than on women. All of the population groups perceived a significant reduction in visits ranging from 0.57 to 0.20, except for women over 81 who increased their visits by 0.37. However, this group consists of only 4 women, representing 1.1% of the whole population; this also happens with the 3 men over 81, which saw the biggest reduction in visits (1.90) but barely represent 0.87% of the population. As such, their results are considered not statistically important, neither here nor when we consider other aspects of use of healthcare resources.

As for the prospective point, the number of visits made by the control group was significantly smaller than that of the case group, both before and after diagnosis. The overall mean stands at 0.81 visits per month, with higher values for older age-groups and women.

When comparing visits before and after diagnosis patient by patient, individually, in primary care, most patients have under 5 visits both before and after diagnosis, with a great number of them staying under 2.5; the distribution is right-skewed. Figure 1 displays an overall reduction in visits from 4,714 to 4,121 (-12%) as seen in the first category (which totals the rest of services). All services with over 100 visits have a strong reduction in visits.

We analysed the ratio of visits in hospital care by age-group and sex, again both prospective and retrospective (Table 5). For the retrospective point, the mean number of visits was 0.72 per month before diagnosis and 0.90 after, with an increase of 0.18 visits that was consistent both for men and women. This consistency arises from the fact that most of these visits were under the programme and thus compulsory for all patients. Most groups displayed a consistent increase in visits from 0.18 to 0.24 except for 66–80 year old women, who saw a reduction of 0.13 visits.

For the prospective point, the overall mean of the control group stood at 0.15 visits per month. The values were higher for older age-groups, and non-significantly higher for women except in the 65–80 and 81 + age-groups. Nonetheless, all population groups in the control populations made less visits to hospital care than their case population counterparts.

When comparing outpatient visits before and after diagnosis patient by patient, the distribution is right-skewed: most patients have under 3 visits. This time around however, a slightly larger number of points is over the red line, showing an increase of visits.

Figure 2 presents a dumbbell plot similar to Fig. 1, now including hospital services (with more than 5 visits made during the treatment period) instead. There was an overall increase in visits from 1,472 to 1,835 (+ 25%). HIV and infectious diseases (UMI) remained the most-visited services, with pneumology, rehabilitation, cardiology, occupational therapy and rheumatology closely following. Most of the services had an increase in visits, except for occupational therapy, neurology, endocrinology and urology.

When comparing the number of hospitalizations by service in the six months before and after diagnosis, there was a 62% reduction in hospitalizations (from 54 to 20). Most of this decrease was observed in admissions related to infectious diseases and HIV services. All but two services (ophthalmology and gynaecology) experienced a reduction in hospitalizations; when there was an increase, it was of only one visit.

When comparing the number of hospital emergencies received by each service, there was a reduction of 24%, from 164 to 124. As expected, the general emergency service received 60% of the visits. Most services did not experience a significant change, except for the general emergency service, obstetrics (which experienced a slight increase) and orthopaedics/traumatology (which experienced a slight reduction).

Comparing the six months before diagnosis with the six months after diagnosis, there was an overall reduction of 32% in the total number of procedures, from 320 to 128. The most common order was that of R00001, chest X-rays, which accounted for 29% of the total orders before diagnosis and 20% after. Second place went to PD00642, liquid-based cytology, which saw the strongest increase at 183%. Behind were chest CT (RA00248) and spine CT (RA00047) scans.

Finally, Table 6 acts as a monetization of the healthcare resources described. These costs are presented for an individual patient in a single month.

Primary care is disaggregated into several groups by service; hospital care, by whether the visit is the first one or successive/others; hospitalizations, by its character in general, surgery or home hospitalizations. Neither emergencies nor procedures are disaggregated. Finally, it presents the total cost of all patients in each group (341 for both pre- and post-diagnosis, 49,078 for control) for six months; then for an individual patient at six months; and finally for an individual patient in a single month.

Comparing the case and control populations, the former is 3.64-fold as expensive against the post-diagnosis population. Comparing pre- and post-diagnosis populations, there is a reduction in primary care costs (16%), hospitalizations (64%), procedures (27%) and emergency care (24%), while there is an increase in hospital care costs (27%), mirroring the results provided earlier in this section.