Protocol for an ambidirectional cohort study on long COVID and the healthcare needs, use and barriers to access health services in a large city in Southeast Brazil

Abstract

Introduction

Post-COVID-19 condition, or syndrome, also known as long COVID, is an infection-associated chronic condition that can develop after a SARS-CoV-2 infection and last at least 3 months to years. Despite representing a high burden for the Unified Health System (SUS), which has affected millions of Brazilians, it has received limited attention in Brazil. Prevalence studies to date have failed to include a broad representation of the population, and there has been insufficient exploration of the impact on people’s lives and the burden of and barriers to accessing health services. This article presents the research protocol for the quantitative component of a mixed methods project to produce evidence to inform SUS’s provision of care for long COVID. The protocol was designed to study long COVID in SUS patients hospitalised for COVID-19 in a large city in Southeast Brazil to capture symptoms and factors associated with the syndrome, effects on quality of life and employment, health needs, use of health services and barriers to accessing necessary healthcare.

Methods and analysis

An ambidirectional cohort study to capture data retrospectively and prospectively from adults previously discharged from SUS hospitals for COVID-19. The study involves up to two telephone surveys with the patients or proxies selected from a sampling plan for population estimates. Survey questions include baseline and follow-up data on demographic, socioeconomic, comorbidities, work status, health-related quality of life, vaccination status, long COVID symptoms, healthcare needs, use and barriers to access. Descriptive and appropriate multivariable analyses will be employed.

Ethics and dissemination

The project was approved by the Research Ethics Committees of participant institutions and by the Brazilian National Research Ethics Commission. All participants provided verbal consent. We plan to publish articles in scientific journals and multimedia resources for SUS professionals and the general population.

Correspondence to Dr Margareth Crisóstomo Portela; [email protected]

STRENGTHS AND LIMITATIONS OF THIS STUDY

• A random sample design that allows for population inferences in a large city.

• Survey administration by telephone enables reaching a geographically distributed population, including those in high social vulnerability and violent areas.

• Capacity to capture information beyond symptoms critical to informing the organisation of public healthcare services for long COVID, including misalignment between the perceived need for services and access/use.

• Follow-up of patients who had COVID-19 in different moments of the pandemic, allowing for explorations about the effects of vaccination and reinfections.

• The study does not allow for inferences about long COVID among those who had mild and moderate COVID-19, nor among extremely socially vulnerable people such as people experiencing homelessness who do not have a telephone.

Introduction

Post-COVID-19 condition, or syndrome, also known as long COVID, is an infection-associated chronic condition that can develop after a SARS-CoV-2 infection and last at least 3 months to years.^{1 2} It can affect anyone exposed to SARS-CoV-2, regardless of age, vaccination status, or the severity of the original symptoms,^{3 4} and has an episodic pattern, which can improve or worsen over time.^{1 5}

Although estimates of the prevalence of long COVID symptoms can vary, studies worldwide point to elevated numbers. Systematic reviews involving more than 1.2 million participants who had COVID-19 in diverse countries indicated that 50.1% remained with at least one symptom for up to a year after the acute infection⁶ and 41.7% had at least one unresolved symptom, and 14.1% could not return to work 2 years after COVID-19 infection.⁷ The most frequent symptoms reported have been fatigue, dyspnoea, muscle and joint pain, cognitive and memory impairments, sleep disorders and psychiatric symptoms.^6–8 Globally, studies have indicated a higher risk of long COVID among female sex individuals, older patients and those with the most severe COVID-19.^{6 7} However, the expansion of studies on broader populations has ratified female sex as a risk factor while indicating the effect of age as nonlinear, with long COVID more likely to be diagnosed between 36 and 50.^{9 10}

Brazil was strongly affected by the pandemic, and long COVID has been studied in convenience samples of different population groups in diverse settings,^11–24 with findings indicating a high prevalence of long COVID, measured as the presence of at least one symptom, and frequent reporting of the onset or persistence of three or more symptoms^{17 18} and a severe impact on the quality of life.¹⁴ The Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística), in a national household sample survey in the first trimester of 2023, estimated that one in four Brazilian adults who had had COVID-19 developed post-COVID syndrome, keeping or not the condition in the moment of the survey.²⁵ Nevertheless, the country still needs more statistically representative studies, primarily focusing on the population assisted by the universal and public Unified Health System (Sistema Único de Saúde—SUS), to understand healthcare needs and support the organisation of the health services network to meet them. Long COVID has received insufficient attention from the health system and policymakers despite potentially representing a high burden and significant challenge for the SUS,^26–29 which covered more than 70% of the COVID-19 hospitalisations in the country,³⁰ provides healthcare to the most vulnerable population groups, and in which specialised care is a ‘bottleneck’.

The protocol presented here is being adopted in studies in the North and Southeast of Brazil, with necessary adjustments due to the different research contexts, including, among other things, varying challenges to accessing data. This article focuses on the study in a large city in Southeast Brazil with the highest number of documented COVID-19 accumulated cases in the country,³¹ corresponding to the quantitative component of a mixed methods project to produce evidence to inform SUS’s provision of care for long COVID.

Objectives of the study

The overarching objective of this research is to study long COVID in patients hospitalised for COVID-19 in the SUS in a large city of Southeast Brazil to understand the characteristics, development and factors associated with the syndrome as well as health needs, use of health services and barriers to accessing necessary healthcare, in order to produce evidence to support the organisation of the health services network to provide appropriate care for people with long COVID.

Specific objectives include (1) to estimate the prevalence of long COVID symptoms and posthospitalisation mortality; (2) to identify factors associated with long COVID; (3) to assess the effects of long COVID on the health-related quality of life and employment status of participants and (4) to characterise long COVID patients’ need for, use of and barriers to accessing health services.

Methods and analysis

Patient and public involvement

A patient-engaged interdisciplinary and international collaboration underpins this mixed methods project. In addition to expertise from lived experience, the research team gathered health researchers with expertise in the local health system and fields including epidemiology, medicine, healthcare improvement, social work, social psychology and community health. Two coauthors live with long COVID, one in Brazil, one international; both members of the Patient-Led Research Collaborative, a group of long COVID patients and patients with associated illnesses such as myalgic encephalomyelitis/chronic fatigue syndrome and postural orthostatic tachycardia syndrome, who are also researchers. They contributed to many aspects of the quantitative component design, such as survey content, instrument design and administration.

Study design and population

We employed an ambidirectional cohort study to capture data retrospectively and prospectively from people aged at least 18 years who were discharged from SUS hospitals for COVID-19 (with a confirmed positive test or clinical diagnosis) from December 2020 to November 2022. Considering the time between hospital discharge and the research team’s contact, the population was stratified into four cohorts: 24, 18, 12 and 6 months postdischarge.

Sampling plan

We opted to focus the study on patients hospitalised during the acute phase of COVID-19 in order to have a frame for the sampling design and achieve a statistical representation of the target population. We adopted a two-stage sampling plan: hospital and patient.

Initially, we considered data from the publicly accessible SUS Hospital Information System (SIH-SUS)—https://datasus.saude.gov.br/transferencia-de-arquivos—which gathers hospital authorisations paid for and refused by SUS and selected the 19 hospitals with the highest volume of hospitalisations due to COVID-19 in the city in 2021, when the peak of hospitalisations due to COVID-19 was observed within the study’s time frame. These hospitals accounted for 76.7% of hospitalisations for COVID-19 in the SUS in the city in the designated period. Next, we endeavoured to obtain the consent of hospitals’ respective management to participate in the study.

Operationally, however, the use of the Severe Acute Respiratory Syndrome Surveillance System (SIVEP-Gripe) with patient identification, which aggregates data collected on severe acute respiratory infection cases in the country, including hospitalisations for COVID-19, was advantageous due to the availability of relevant data on hospitalisations and contact details for reaching patients. We decided to use it from then on, with the consent of the Health Surveillance Secretariat of the city’s Municipal Health Secretariat and the Research Coordination Department of the respective State Health Secretariat. The comparison of SIH-SUS with SIVEP-Gripe (https://opendatasus.saude.gov.br/nl/dataset) data revealed inconsistencies, which led us to use the number of survivors from each hospital in the SIVEP-Gripe to measure the sample size in the selected hospitals.

As is common in selection with probability proportional to size, some units with a large size were included in the sample with certainty. In this case, these units (hospitals) became selection strata, and the patients became the primary sampling units. Before their selection, patients were stratified by cohort and selected by reverse sampling.^32–34 Patients were registered in random order and sequentially searched within each cohort (24, 18, 12 and 6 months after hospital discharge).

The total sample size was calculated to estimate a minimum proportion of 3% (P_min=0.03), with a relative error of no more than 0.5% at a significance level of 5%, which implies that this proportion will have a 95% CI ranging from 1.5% to 4.5%.

According to Cochran³⁵ and assuming simple random sampling without replacement (SRSWOR), the sample size needed to estimate proportions equal to or greater than P_min with a relative error no more significant than d_r at a confidence level of 1-α is given by:

(a) nSRSWOR=z∝/22dr2×1−PminPmin

where $z_{\propto /2}$ is the quantile (1–α/2) of the standard normal distribution. The application of this expression leads to n_SRSWOR=346.

However, the sample selected is clustered by hospital, and Pessoa and Silva³⁶ recommend multiplying the sample size obtained by expression (a) by an estimate of the sampling plan effect (SPE) for the sizing variable. An SPE of 1.4 was arbitrated for use in sizing the sample since most of the hospitals were undoubtedly included in the sample. Thus, the final sample size of patients became n_SRSWOR×SPE = 346×1.4=484.

The total sample size was allocated between the hospitals in proportion to the size of the hospital (ie, the number of surviving patients), ensuring a minimum of five patients per hospital. The sample size of the hospitals was allocated between the cohorts in proportion to the number of survivors in each cohort.

The sample weights of the patients interviewed were initially calculated as the inverse of the product of the probabilities of inclusion in each stage of the sample

(b) ${\displaystyle {\mathrm{W}}_{\mathrm{j}\mathrm{k}\mathrm{i}}=\frac{1}{\mathrm{P}\left({\mathrm{H}}_{\mathrm{i}}\right)}\times \mathrm{P}({\mathrm{P}}_{\mathrm{j}\mathrm{k}}|{\mathrm{H}}_{\mathrm{i}})}$

$W_{jki}$ is the natural design weight of patient j, cohort k and hospital i.

${\displaystyle \mathrm{P}\left({\mathrm{H}}_{\mathrm{i}}\right)}$ is the probability of including hospital i in the sample.

${\displaystyle P(P_{jk}|H_i)}$ of patient j in cohort k given the selection of hospital i.

As only 16 of the 19 hospitals—10 municipal, two state, two federal and two university—agreed to participate in the study, it was necessary to introduce a non-response adjustment to mitigate its effects. This adjustment was applied to the natural weight of the hospital, P(H_i), and the natural weight of the patient given the hospital selection, P(P_jc | H_i). Thus, the adjusted probability of inclusion, represented by ${\displaystyle \mathrm{P}({\mathrm{H}}_{\mathrm{i}}^{\ast })}$ , was recalculated, changing the sample size of hospitals from 19 to 16 and maintaining the size measures of the hospitals; in other words:

(c) ${\displaystyle {\displaystyle {\mathrm{P}(\mathrm{H}}_{\mathrm{i}}^{\ast }\text{)=16}\times \frac{\sum _{\mathrm{i}}{\mathrm{M}}_{\mathrm{i}}}{{\mathrm{M}}_{\mathrm{i}}}}}$

M_i is the measure of the size of hospital i (total surviving patients).

As it was not always possible to achieve the expected sample size of patients, the adjusted probability of patients from each hospital and cohort, conditional on hospital selection, represented by ${\displaystyle \mathrm{P}({\mathrm{P}}_{\mathrm{j}\mathrm{k}}^{\ast }|{\mathrm{H}}_{\mathrm{i}}^{\ast })}$ , was recalculated as the ratio between the number of patients interviewed and the number in SIVEP-Gripe, both per hospital and cohort, that is,

(d) $\mathrm{P}\left({\mathrm{P}}_{\mathrm{j}\mathrm{k}}^{\mathrm{*}}\right|{\mathrm{H}}_{\mathrm{i}}^{\mathrm{*}})=\frac{{\mathrm{n}}_{\mathrm{j}\mathrm{k}}}{{\mathrm{N}}_{\mathrm{j}\mathrm{k}}}$

${\mathrm{n}}_{\mathrm{j}\mathrm{k}}$ is the number of patients interviewed (sample size) in cohort k of hospital i and

${\mathrm{N}}_{\mathrm{j}\mathrm{k}}$ is the number of patients (population size) in SIVEP-Gripe in cohort k at the hospital i.

Thus, the patient’s weight, adjusted for non-response, can be described by

(e) ${\displaystyle {\mathrm{W}}_{\mathrm{j}\mathrm{k}\mathrm{i}}^{\ast }=\frac{1}{\mathrm{P}\left({\mathrm{H}}_{\mathrm{i}}^{\ast }\right)}\times \mathrm{P}({\mathrm{P}}_{\mathrm{j}\mathrm{k}}^{\ast }|{\mathrm{H}}_{\mathrm{i}}^{\ast })}$

As sex and age are important variables in data analysis, we decided to calibrate the adjusted sample weights $W_{jki}^{*}$ using a post-stratification calibration between the population totals (obtained from SIVEP-Gripe – https://opendatasus.saude.gov.br/nl/dataset) and the totals estimated with the adjusted weights ${\mathrm{W}}_{\mathrm{j}\mathrm{k}\mathrm{i}}^{\mathrm{*}}$ .

The poststrata were defined by cohort, sex and age group defined in complete years as 18–39, 40–49, 50–59, 60–69, 70–79 and 80 or more.

The expression that represents the calibrated weight, represented by $W_{jki}^c$ , is given by:

(f) ${\mathrm{W}}_{\mathrm{j}\mathrm{k}\mathrm{i}}^{\mathrm{c}}={\mathrm{W}}_{\mathrm{j}\mathrm{k}\mathrm{i}}^{\mathrm{*}}\times \frac{{\mathrm{N}}_{\mathrm{p}}}{\widehat{{\mathrm{N}}_{\mathrm{p}}}}\forall \mathrm{j}\mathrm{k}\in \mathrm{p}$

${\mathrm{N}}_{\mathrm{p}}$ is the number of poststratum p patients in SIVEP-Gripe

$\widehat{{\mathrm{N}}_{\mathrm{p}}}$ is the estimated number of post-stratum p patients, estimated with the weight adjusted for non-response.

Online supplemental material 1 shows the population totals by post-stratum, corresponding to the COVID-19 discharges of surviving patients in SUS hospitals in the city between December 2020 and November 2022.

Data collection

The SIVEP-Gripe identified database, from which we obtained the randomly sorted lists of potential study participants, is kept under the strict control of two study team members, the general coordinator and the field coordinator. All potential participants were given a code to anonymise the data. Partial lists with identification and contact details of the selected individuals were distributed among the interviewers to contact and invite them to participate in the study.

The study involves telephone interviews with patients discharged from hospital due to COVID-19. In the event of the selected patients’ death or difficulties participating, we invited people close to them (eg, spouse/widow, daughter/son, or caregiver) who could answer the questions on their behalf.

Given that the telephone numbers available on SIVEP-Gripe usually correspond to contact information about the patient during hospitalisation, the interviewers’ first effort was to obtain the telephone numbers for direct contact with the selected patients or identify alternative respondents if they could not do so. The selected patients and other respondents were contacted by telephone, informed of the nature and objectives of the research and invited to participate after clarifying what was expected of their participation. Material about the research was produced and sent to potential study participants by email or WhatsApp. Consent to take part in the study was obtained verbally and audio-recorded.

Each patient’s follow-up includes up to two interviews. The first was held around 6, 12, 18 and 24 months after discharge from COVID-19 hospitalisation, and the second was programmed to be held approximately 6 months later. We decided not to attempt to contact living participants for a second interview if the participant had declared that they did not self-perceive as having long COVID and simultaneously did not report at the time of the first interview at least three selected long COVID symptoms (fatigue, postexertional malaise, breathlessness, cognitive impairment, joint pain, sleep alterations, numbness, dizziness, depression, anxiety, muscle pain, palpitations, vision problems, difficulty walking and headache), nor needed health services for new or deteriorated health problems after COVID-19.

Interviews were scheduled for the day and time of the patient’s convenience and held by telephone or video platform. A researcher applies the questionnaire on the RedCap Platform; the patients do not self-complete questionnaires. Initially, the possibility of holding some interviews in person was envisaged, respecting the convenience of the participants, but no participant opted for in-person interviews. Also, we occasionally split interviews into two different stages for the convenience of participants (eg, to accommodate their symptoms).

Variables of interest and data collection instrument

A structured questionnaire was constructed (online supplemental material 2),³⁷ including questions for the first interview to compose a baseline relating to the pre-COVID-19 illness period. Baseline data include demographic, socioeconomic, lifestyle, health (comorbidities), work status, health-related quality of life variables and the participant’s vaccination status before COVID-19. The other questions captured their clinical progress since discharge from hospital/last interview—persistent symptoms, emergence or deteriorated comorbidities, perceived care needs, care received, barriers to accessing health services, current employment situation and income, health-related quality of life, COVID-19 reinfections, current vaccination status and death.

The occurrence of a participant’s death is recorded at the beginning of the interview, with the date and information on whether or not it was linked to post-COVID-19 vulnerability; in this event, follow-up questions are not asked. In the case of death between discharge and the first interview of the study, data on quality of life and employment status are not registered at baseline either.

The demographic and socioeconomic variables included are date of birth (to confirm the one available in SIVEP-Gripe), gender, race and ethnicity, schooling, per capita household income and employment status before COVID-19. Data on comorbidities and health-related quality of life before COVID-19 are also obtained to form the ‘baseline’ on the participant’s health situation.

Health-related quality of life is assessed using the EuroQol EQ5D-5L instrument,³⁸ which covers five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. The questions are asked in the past to capture the baseline pre-COVID-19 moment and, in the present, so that the participants may indicate their current state of health in each dimension.

The outcome variables relating to the participants’ health status include persistent and new symptoms and their effects on work and health-related quality of life, measured using the EuroQol EQ5D-5L. Survival after discharge is another outcome recorded.

Additionally, regarding the need for and use of health services for long COVID care, the questionnaire includes variables capturing whether the participant needed and had access to specified services (primary healthcare appointments, hospitalisation/emergency, rehabilitation, specialist appointments, mental healthcare, complementary medicine, pharmacy, laboratory tests, imaging tests) and barriers in accessing the SUS. Also recorded are out-of-pocket cost patients or their families had with tests, medicines and healthcare professional visits in the previous month.

A qualified team, including senior researchers and master’s and doctoral students, was involved in data collection. Through training and periodic meetings, the team worked towards alignment and consistency in the questionnaire application.

Of note is that the study is also designed to include information pertinent to the index hospitalisations of the sampled patients, for which secondary data from SIVEP-Gripe itself will be used. Of particular interest are variables indicating the hospitalisation severity level—invasive or non-invasive ventilatory support and ICU use—to be explored as explanatory variables of outcomes of interest.

Analyses

The data obtained should help characterise the patients, their clinical progress and their access to and use of health services due to long COVID during the observation period. Descriptive statistics will include absolute and relative frequencies for categorised variables and mean, SD and quartiles for numerical variables. Comparisons of the means of numerical variables before and after COVID-19 will be made, and associations between categorised variables will be explored using the χ² test .

Logistic, linear and survival (Cox) regression models will be employed to identify factors associated with dependent variables of interest, considering the variable type.

In all analyses, we will account for the sample design variables—selection strata, primary sampling units and sample weights—employing procedures of the SAS statistical package oriented towards complex survey data.

Finally, triangulating these data with those from the qualitative component of our study, which included interviews with managers, health professionals and patients living with long COVID, will allow elaborating viable recommendations relevant to the context.

Figure 1 provides a summary of the research protocol.

Figure 1. Summary of the protocol for an ambidirectional cohort study on long COVID and the healthcare needs, use, and barriers to access health services in a large city in Southeast Brazil. SUS, Sistema Único de Saúde.

Discussion

The study recruited the participants from 22 November 2022 to 30 August 2023 and closed a sample of 651 patients. The second wave of interviews with a part of the same sample and the data collection for incomplete hospitalisation registers were closed in March and April 2024, respectively. Results have not been released yet. The unidentified databases are expected to be released on a public platform once the results have been published.

Below, we discuss the study design’s strengths and limitations, including assessments made in the research operationalisation relevant to other studies in the Brazilian context and, eventually, to similar contexts worldwide.

Population of interest and sample

The first element of the study design that required much thought was defining the population to be targeted and selecting the sample. As the study aimed to provide evidence for healthcare in the SUS, it was natural to focus on the population that uses health services in the public health system. There were doubts about focusing on the population who had COVID-19 regardless of the acute condition’s severity or focusing on long COVID among those requiring hospitalisation. Recruiting people assisted in primary healthcare units could be a good option. Still, three reasons made us discard it: irregular data on previous COVID-19 diagnoses due, among other reasons, to the lack of tests, which was critical in some moments of the pandemic in Brazil,³⁹ and the capacity of the units to register results; incompleteness and reliability of patients’ data, including contact information; and lack of a single patient registry for all units, without which research logistics would be even more challenging. These conditions would not allow for an appropriate sampling design.

The choice to focus the study on patients who had been hospitalised due to COVID-19 in the SUS, therefore, was pragmatic, given the availability of records of these patients, despite recognising that the choice imposed the limit of excluding people with long COVID following mild or moderate acute COVID-19 cases, which did not lead to hospitalisation. The idea of conducting a study statistically representative of the population prevailed. However, it circumscribed the population to a subpopulation, in which the expectation of long COVID occurrence would be higher and, in part, could even be confused with complications due to extended stays in intensive care units. Anyway, we assumed that looking at the ‘tip of the iceberg’ would bring important information about the syndrome in the Brazilian context once we are cautious about feasible interpretations and conclusions.

Another challenge that surprised the study team was selecting the admissions register to be considered, given the significantly inconsistent number of admissions per hospital recorded in SIH-SUS and SIVEP-Gripe. The finding points to the need to improve the quality of fundamental data for epidemiological and health service surveillance, especially in a health emergency, which may have contributed to poor records. Duplicated records, flows and even database management entities are problems that need to be tackled to streamline the use of resources and improve data quality. We opted for the SIVEP-Gripe as the population registry for sample selection, considering that the second hospital in volume of COVID-19 hospitalisations in the city had few records of COVID-19 admissions in the SIH-SUS. The availability of variables relating to the testing and diagnostic confirmation of cases in the SIVEP-Gripe, a database designed to monitor severe acute respiratory syndromes, was also crucial.

Telephone interviews

The study design allowed for telephone, video or face-to-face interviews, with all initial contacts by telephone, using the number available on SIVEP-Gripe. There was a great deal of loss in making initial contact with the patients selected for the sample, who were replaced per a randomly sorted list. These losses were primarily due to wrong numbers registered or no longer being valid. Since the available numbers are patient contact numbers given to hospitals for information about them, another level of loss occurred between contact with the bearer of the available telephone number and the patients themselves. Some reasons are resistance to providing the number, lack of a closer relationship with the patients and knowledge about their whereabouts or even anticipating the patient’s refusal to participate in the study. When the study was conceived, we assessed that the balance between obtaining data provided by someone close to the patient and losing the data favoured the first alternative, given the possibility of studying the event of death after hospital discharge and possibly cases with more significant functional impairment after COVID-19. This procedure was also done in rare instances where the patient had no telephone, and the contact was a close family member directly following the patient’s health situation. Several contacts were facilitated by the WhatsApp application, through which research material was sent. The impossibility of contacting ‘landline’ telephones was a constant, but the availability of mobile telephone numbers was widespread.

Once we contacted the patient or, in the event of the patient’s death or inability to answer the questionnaire, someone close to the patient who could answer the questionnaire on their behalf, we obtained their authorisation to participate in the study and scheduled the interview. Fewer potential participants were lost at this stage. Almost all participants opted for telephone interviews.

Despite applying a lengthy questionnaire, especially in the first interview, telephone interviews effectively reached a geographically distributed population, including those in areas of high social vulnerability and violence.³⁷ They could also reliably capture the data provided by the respondents, as has already been shown in other studies.^{40 41} However, we must recognise the high likelihood of recall bias in some baseline questions, among which we underline those about health-related quality of life. In contrast to studies whose recruitment occurs in health service settings, we also underscore the possibility of capturing participants with or without health needs who are not assisted in these services. However, the potentially less circumventable bias concerns the non-capture of highly vulnerable people, including people in street situations, who neither had a telephone nor contacts to offer adequate information.

COVID-19 prevalence

Another point worth highlighting in the study is the difficulty in estimating the prevalence of long COVID, although this objective had been initially set. Long COVID does not have a biomarker, and the diagnosis is established based on the persistent symptoms after the acute COVID-19, with no other identified causes. Given the wide range of symptoms and the concern about administering a lengthy questionnaire to participants who could have limitations imposed by the long COVID condition, we opted to include a parsimonious list of symptoms in the study, covered by 30 questions.^{42 43} We also asked whether the participants perceived themselves having long COVID or whether they had been diagnosed with long COVID by a health professional.

As with other research, this study allows us to estimate the prevalence of long COVID symptoms and cross-reference these symptoms with the individual’s perception of having long COVID and the presence of pre-existing comorbidities. The inclusion of comorbidities enables hypothesis testing about long COVID risk factors. Also, this study allows us to capture the participants’ perception of using health services for health problems identified as new or deteriorated after COVID-19.

The study sample design’s potential for making population estimates is undoubtedly a strength compared with most studies conducted from purposive samples. However, the anticipated limitation of not including long COVID cases in people not hospitalised during the acute phase of COVID-19 persists. Sensitivity analyses on possible long COVID distribution settings among non-hospitalised people, partly fed by findings regarding hospitalised patients, could mitigate this limitation. Data from the FAIR Health’s repository of 78 252 patients diagnosed with LC (long COVID), for example, indicated that 81.6% of women, compared with 67.5% of men, had not had a COVID-19 hospitalisation.⁹

Study coverage

The effort to measure long COVID in the population of patients hospitalised for COVID-19 in the SUS in the city is relevant to give visibility to the problem, so that evidence-based actions can be developed at the primary and secondary healthcare levels to achieve care centred on people living with post-COVID-19 conditions. However, this study also aims to map the healthcare needs, use and barriers to accessing health services in the SUS, which is critical and less explored in studies on long COVID. We expect that the study will also make an original contribution regarding the impact of long COVID on working life and income in the city and on spending on medication, appointments and tests not obtained from the SUS. Moreover, the research will add information on long COVID-related risk factors, on the effects of long COVID on health-related quality of life, on factors associated with the survival of COVID-19 patients after hospital discharge, and on COVID vaccination’s role in long COVID.

As a final assessment, we would like to highlight the research team’s perception that the methodological choices, although with expressed limitations, allow for a valuable and pragmatic approach to capture the population of interest regarding representativeness and the potential for acceptable biases concerning the available data reality.

Ethics and dissemination

The project was submitted to and approved by the Research Ethics Committees of ENSP/Fiocruz (CAAE 57680922.3.0000.5240), of the city’s Municipal Health Secretariat (CAAE 57680922.3.3001.5279), and of one of the coparticipant hospitals (CAAE 57680922.3.3003.5257), as required by its Management. Given the international collaboration and funding, it was also submitted and approved by the National Research Ethics Commission (CONEP) (CAAE 57680922.3.0000.5240). All participants provided oral consent.

With regard to dissemination, we plan to publish at least four articles in scientific journals focused on each of the following: long COVID symptoms; healthcare needs, use and barriers; analyses of survival post- discharge; effects of long COVID on health-related quality of life; and effects of long COVID on employment and income. We also plan to produce videos and informative materials oriented towards healthcare managers and professionals and the general population.

MCP and MM received productivity fellowships from the Brazilian National Council of Research and Technological Development (CNPq). The authors are grateful to EuroQol for permission to access and use the instrument in the survey.

Ethics statements

Patient consent for publication

Not applicable.

Footnote

X @mcportela1811

Contributors MCP: conceptualisation, data curation, funding acquisition, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. MTLdV, CLTdA, TLMA, CdAA, LS, ES: conceptualisation, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. SMLL: conceptualisation, funding acquisition, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. BdNC: conceptualisation, funding acquisition, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. MM: conceptualisation, methodology, writing (review and editing, agreement to be accountable for all aspects of the work. MB: conceptualisation, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. E-LA: conceptualisation, funding acquisition, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. MBR: conceptualisation, funding acquisition, methodology, writing (original draft preparation, review, editing and final approval), agreement to be accountable for all aspects of the work. MCP is responsible for the overall content as guarantor.

Funding MR received a grant from the Harvard University Lemann Brazil Research Fund–https://news.harvard.edu/gazette/story/newsplus/2022-lemann-brazil-research-fund-awardees-announced/–and MCP and SMLL received another grant from the Programa de Fomento ao Desenvolvimento Científico e Tecnológico, ENSP/Fiocruz (ENSP-024-FIO-21-2-12)–https://informe.ensp.fiocruz.br/secoes/secao/45085. The funders played no role in the study design and development, publication decision, or manuscript preparation.

Competing interests None declared.

Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

Provenance and peer review Not commissioned; externally peer-reviewed.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Greenhalgh T, Sivan M, Perlowski A, et al. Long COVID: a clinical update. Lancet 2024; 404: 707–24. doi:10.1016/S0140-6736(24)01136-X

² Centers for Disease Control and Prevention. Post-covid conditions. 2024. Available: https://www.cdc.gov/coronavirus/2019-ncov/long-term-effects/index.html [Accessed 27 Sep 2024 ].

³ World Health Organization. Post covid-19 condition (long covid). Available: https://www.who.int/europe/news-room/fact-sheets/item/post-covid-19-condition [Accessed 27 Sep 2024 ].

⁴ Al-Aly Z, Bowe B, Xie Y. Long COVID after breakthrough SARS-CoV-2 infection. Nat Med 2022; 28: 1461–7. doi:10.1038/s41591-022-01840-0

⁵ Ministério da Saúde. Nota Técnica 57/2023 - DGIP/SE/MS. Atualizações acerca das “condições pós-COVID” no âmbito do Ministério da Saúde. Brasil Ministério da Saúde; 2023. Available: https://bvsms.saude.gov.br/bvs/publicacoes/nota_tecnica_n57_atualizacoes_condicoes_poscovid.pdf [accessed 27 Sep 2024 ]

⁶ Zeng N, Zhao Y-M, Yan W, et al. A systematic review and meta-analysis of long term physical and mental sequelae of COVID-19 pandemic: call for research priority and action. Mol Psychiatry 2023; 28: 423–33. doi:10.1038/s41380-022-01614-7

⁷ Rahmati M, Udeh R, Yon DK, et al. A systematic review and meta-analysis of long-term sequelae of COVID-19 2-year after SARS-CoV-2 infection: A call to action for neurological, physical, and psychological sciences. J Med Virol 2023; 95: e28852. doi:10.1002/jmv.28852

⁸ Alkodaymi MS, Omrani OA, Fawzy NA, et al. Prevalence of post-acute COVID-19 syndrome symptoms at different follow-up periods: a systematic review and meta-analysis. Clin Microbiol Infect 2022; 28: 657–66. doi:10.1016/j.cmi.2022.01.014

⁹ FAIR Health. White paper - Patients diagnosed with Post-COVID conditions: an analysis from private healthcare claims using the official ICD-10 diagnostic code. 2022. Available: https://digirepo.nlm.nih.gov/master/borndig/9918504887106676/9918504887106676.pdf [Accessed 27 Sep 2024 ].

¹⁰ Centers for Disease Control and Prevention (CDC). Long covid: household pulse survey. 2024. Available: https://www.cdc.gov/nchs/covid19/pulse/long-covid.htm

¹¹ Rosa RG, Cavalcanti AB, Azevedo LCP, et al. Association between acute disease severity and one-year quality of life among post-hospitalisation COVID-19 patients: Coalition VII prospective cohort study. Intensive Care Med 2023; 49: 166–77. doi:10.1007/s00134-022-06953-1

¹² Busatto GF, de Araújo AL, Duarte AJ da S, et al. Post-acute sequelae of SARS-CoV-2 infection (PASC): a protocol for a multidisciplinary prospective observational evaluation of a cohort of patients surviving hospitalisation in Sao Paulo, Brazil. BMJ Open 2021; 11: e051706. doi:10.1136/bmjopen-2021-051706

¹³ Lapa J, Rosa D, Mendes JPL, et al. Prevalence and Associated Factors of Post-COVID-19 Syndrome in a Brazilian Cohort after 3 and 6 Months of Hospital Discharge. Int J Environ Res Public Health 2023; 20: 848. doi:10.3390/ijerph20010848

¹⁴ Bonifácio LP, Csizmar VNF, Barbosa-Júnior F, et al. Long-Term Symptoms among COVID-19 Survivors in Prospective Cohort Study, Brazil. Emerg Infect Dis 2022; 28: 730–3. doi:10.3201/eid2803.212020

¹⁵ Nakayama LF, Urias MG, Gonçalves AS, et al. Post-discharge follow-up of patients with COVID-19: A Brazilian experience. SAGE Open Med 2022; 10: 20503121221096602. doi:10.1177/20503121221096602

¹⁶ Titze-de-Almeida R, da Cunha TR, dos Santos Silva LD, et al. Persistent, new-onset symptoms and mental health complaints in Long COVID in a Brazilian cohort of non-hospitalized patients. BMC Infect Dis 2022; 22: 133. doi:10.1186/s12879-022-07065-3

¹⁷ de Miranda DAP, Gomes SVC, Filgueiras PS, et al. Long COVID-19 syndrome: a 14-months longitudinal study during the two first epidemic peaks in Southeast Brazil. Trans R Soc Trop Med Hyg 2022; 116: 1007–14. doi:10.1093/trstmh/trac030

¹⁸ Marra AR, Sampaio VS, Ozahata MC, et al. Risk factors for long coronavirus disease 2019 (long COVID) among healthcare personnel, Brazil, 2020-2022. Infect Control Hosp Epidemiol 2023; 44: 1972–8. doi:10.1017/ice.2023.95

¹⁹ da Costa e Silva GR, Moura WÉA, dos Santos KC, et al. Long-Term Symptoms after Mild Coronavirus Disease in Healthy Healthcare Professionals: A 12-Month Prospective Cohort Study. IJERPH 2023; 20: 1483. doi:10.3390/ijerph20021483

²⁰ da Silva MRT, Costa AP, da Luz AA, et al. From Acute Phase to Long COVID: A Cross-Sectional Study of the Epidemiological Profile and Clinical Evaluation of SARS-CoV-2 Infection in Employees at a Pediatric Hospital. J Clin Med 2023; 12: 4325. doi:10.3390/jcm12134325

²¹ Feter N, Caputo EL, Leite JS, et al. Prevalence and factors associated with long COVID in adults from Southern Brazil: findings from the PAMPA cohort. Cad Saude Publica 2023; 39: e00098023. doi:10.1590/0102-311XEN098023

²² Laskovski L, Felcar JM, Fillis MMA, et al. Risk factors associated with limited functional status among out-of-hospital patients 30 days and one year after a diagnosis of COVID-19: a cohort study. Sci Rep 2023; 13: 3584. doi:10.1038/s41598-023-30674-0

²³ Talhari C, Criado PR, Castro CCS, et al. Prevalence of and risk factors for post-COVID: Results from a survey of 6,958 patients from Brazil. An Acad Bras Cienc 2023; 95: e20220143. doi:10.1590/0001-3765202320220143

²⁴ da Silva NS, de Araújo NK, Dos Santos KA, et al. Post-Covid condition and clinic characteristics associated with SARS-CoV-2 infection: a 2-year follow-up to Brazilian cases. Sci Rep 2023; 13: 13973. doi:10.1038/s41598-023-40586-8

²⁵ Instituto Brasileiro de Geografia e Estatística (IBGE). Pesquisa Nacional por Amostra de Domicílios Contínua 2023. Covid-19 2023 [ Instituto Brasileiro de Geografia e Estatistica ]. 2024. Available: https://biblioteca.ibge.gov.br/visualizacao/livros/liv102087_informativo.pdf

²⁶ Aiyegbusi OL, Hughes SE, Turner G, et al. Symptoms, complications and management of long COVID: a review. J R Soc Med 2021; 114: 428–42. doi:10.1177/01410768211032850

²⁷ Halpin SJ, McIvor C, Whyatt G, et al. Postdischarge symptoms and rehabilitation needs in survivors of COVID-19 infection: A cross-sectional evaluation. J Med Virol 2021; 93: 1013–22. doi:10.1002/jmv.26368

²⁸ Razai MS, Al-Bedaery R, Anand L, et al. Patients’ Experiences of “Long COVID” in the Community and Recommendations for Improving Services: A Quality Improvement Survey. J Prim Care Community Health 2021; 12: 21501327211041846. doi:10.1177/21501327211041846

²⁹ Rushforth A, Ladds E, Wieringa S, et al. Long Covid - The illness narratives. Soc Sci Med 2021; 286: 114326. doi:10.1016/j.socscimed.2021.114326

³⁰ Portela MC, Martins M, Lima SML, et al. COVID-19 inpatient mortality in Brazil from 2020 to 2022: a cross-sectional overview study based on secondary data. Int J Equity Health 2023; 22: 238. doi:10.1186/s12939-023-02037-8

³¹ Ministério da Saúde. COVID-19 no Brasil. Brasil: Ministério da Saúde; 2024. Available: https://infoms.saude.gov.br/extensions/covid-19_html/covid-19_html.html [accessed 27 Sep 2024 ]

³² Haldane JBS. On a method of estimating frequencies. Biometrika 1945; 33: 222–5. doi:10.1093/biomet/33.3.222

³³ Vasconcellos MTL, Silva PLN, Anjos LA. Sample design for the Nutrition, Physical Activity and Health Survey (PNAFS). Niterói Rio Janeiro Brazil Estatística 2013; 65: 47–61. doi:10.1590/0102-311X00037221

³⁴ Vasconcellos MTL de, Silva PL do N, Szwarcwald CL. Sampling design for the World Health Survey in Brazil. Cad Saúde Pública 2005; 21: S89–99. doi:10.1590/S0102-311X2005000700010

³⁵ Cochran WG. Sampling techniques. 3rd edn. New York: John Wiley & Sons, 1977.

³⁶ Pessoa DG, Silva PLN. Análise de dados amostrais complexos. São Paulo: Associação Brasileira de Estatística, 1998.

³⁷ Caldas B, Portela M, Stelson E, et al. Promoting equity, diversity, and inclusion in surveys: insights from a patient-engaged study to assess long COVID health-care needs in Brazil. J Clin Epidemiol 2024; 173: 111423. doi:10.1016/j.jclinepi.2024.111423

³⁸ EuroQol Group. EQ-5D-5L. Available: https://euroqol.org/eq-5d-instruments/eq-5d-5l-about/ [Accessed 27 Sep 2024 ].

³⁹ Barberia LG, de P Moreira N, Kemp B, et al. Evaluation of the effectiveness of surveillance policies to control the COVID-19 pandemic in São Paulo, Brazil. Glob Health Res Policy 2022; 7: 27. doi:10.1186/s41256-022-00260-4

⁴⁰ Marcantonio ER, Michaels M, Resnick NM. Diagnosing delirium by telephone. J Gen Intern Med 1998; 13: 621–3. doi:10.1046/j.1525-1497.1998.00185.x

⁴¹ Wager J, Barth F, Stahlschmidt L, et al. Testing the validity of telephone interviews to assess chronic pain in children and adolescents: A randomized cross-over trial. Eur J Pain 2017; 21: 1707–16. doi:10.1002/ejp.1081

⁴² National Institute for Health and Care Excellence (NICE), Royal College of General Practitioners (RCGP), Scottish Intercollegiate Guidelines Network (SIGN). COVID-19 rapid guideline: managing the long-term effects of COVID-19. 2024. Available: https://www.nice.org.uk/guidance/ng188 [Accessed 27 Sep 2024 ].

⁴³ Munblit D, Nicholson T, Akrami A, et al. A core outcome set for post-COVID-19 condition in adults for use in clinical practice and research: an international Delphi consensus study. Lancet Respir Med 2022; 10: 715–24. doi:10.1016/S2213-2600(22)00169-2