Lessons Learned: Defining the Right Specifications for Diagnostics in Outbreak Situations

Abstract

Introduction

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic significantly changed the development and use of infectious disease diagnostics, emphasizing the need for rapid implementation, access equity, and revealing challenges in evaluation and licensing from unclear guidelines.

Methods

An international expert panel was convened by Roche Diagnostics, Fraunhofer Institute for Translational Medicine and Pharmacology (Immunology, Infection and Pandemic Research division), and Ludwig Maximilians University to address these challenges. The panel aimed to develop parameters for designing minimal viable products and consequence-centered diagnostic strategies, focusing on the broader implications of test results for public health interventions and disease control rather than individual test performance.

Results

Key findings included the need to tailor target product profiles to outbreak scenarios, considering socioeconomic factors and outbreak phases. Prioritizing rapid, cost-effective, and widely accessible tests, even at the expense of lower sensitivity was highlighted. Using such tools was effective during the SARS-CoV-2 pandemic because they were wider-reaching and identified more cases, especially in low-/middle-income countries where highly sensitive molecular tests had limited availability.

Conclusion

Modeling different testing strategies allows outbreak control programs to quantify the impact of trade-offs between accessibility, affordability, and speed of diagnosis in different outbreak settings.

Key Summary Points

Why carry out this study?

The 2019–2020 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic resulted in the rapid development of diagnostic tests and strategies, highlighting the need for defined parameters for minimal viable products and consequence-centered approaches across disease scenarios.

Establishing such parameters is essential to ensure rapid implementation and equitable access to diagnostic tools.

While parameters are often based on expert opinion, this study builds on expert panel outputs by applying a modeling approach to evaluate different diagnostic strategies in a polio-like disease outbreak.

What was learned from the study?

During an outbreak, diagnostic test development should prioritize speed, cost-effectiveness, and availability—particularly in low- and middle-income countries—over maximal sensitivity.

Modeling a range of testing strategies may enable outbreak control programs to quantify and balance trade-offs between accessibility, affordability, and diagnostic speed across different outbreak settings.

Introduction

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic had a significant impact on the development and implementation of infectious disease diagnostics, especially in low- and middle-income countries (LMICs) [1, 2]. Tests have been developed not only at an unprecedented speed and scale, but they have also been used in unexpected settings such as a public health tool to ensure a safe environment for reopening of schools, workplaces, and for mass gatherings [2–5]. However, the licensing and evaluation strategies for diagnostic infectious disease tests have faced challenges due to the large number of new tests on the market and the need for rapid implementation for individual patient care and for public health measures [3, 6]. In addition, clear guidelines on what information a test system should provide to be useful are still lacking.

It is now necessary to reflect on the experiences and determine what can be learned from the testing in order to improve our approach to developing and utilizing tests. This will ultimately lead to more effective diagnostic strategies for infectious disease outbreaks in the future.

To address these questions, an international expert panel was convened with the objective of defining parameters for minimal viable products (MVPs) and developing consequence-centered approaches for diagnostic products and strategies, depending on selected possible disease scenarios. The panel aimed to take a holistic perspective, in order to understand the broader implications of test results and how diagnostic outcomes interact with their subsequent consequences.

Methods

The expert panelists were selected based on their geographic location (to ensure coverage of all major global regions) and their expertise. All panelists had extensive experience in infectious diseases, public health, and epidemiology. They met in person at a workshop organized by a consortium consisting of Roche Diagnostics, Fraunhofer Institute for Translational Medicine and Pharmacology (Immunology, Infection and Pandemic Research division), and Ludwig Maximilians University (LMU) in Lisbon in May 2023.

During the workshop, panelists provided feedback and statements that were recorded in a protocol, which then served as the foundation for this publication. Specifically, the panelists pinpointed key characteristics that should be taken into account when defining outbreak scenarios, especially for LMICs. Considering socioeconomic factors and the different phases of an outbreak, a process was identified to generate tailored target product profiles (TPPs) for diagnostics that specifically address the needs of each scenario. Disagreements were settled through discussion until a consensus was achieved. From these discussions, we generated a blueprint outlining important parameters to be considered during the development of TPPs for MVPs, which was considered essential and could be used for modeling the device’s impact. Additionally, several testing strategies were modeled to support experts’ opinions on the performance of a diagnostic test at a public health level during an outbreak.

The impact of various testing scenarios in a defined outbreak scenario was modeled using a previously described model. No real patient data or samples were used. Ethical approval was, therefore, not required. RocheChat, an artificial intelligence-based conversational agent built on a large language model, was used to summarize panel discussions and rephrase sections of this manuscript. Subsequently, all rephrased sections were reviewed and adjusted as necessary to ensure the content remained accurate and unaltered.

Results

Definition of a Generic Outbreak Scenario

The characteristics that should be considered when constructing scenarios need to be defined, and the reasoning behind their significance provided (Table 1). It is important to emphasize that multiple factors have the potential to influence the outcomes of diagnostic and public health strategies. This comprehensive approach will facilitate a more rigorous and informative evaluation of the scenarios and their implications.

Table 1.

Relevant characteristics for outbreak scenario definition

Characteristic	Rationale for why this is important
Mode and kinetics of transmission	It has a significant impact on the spread (e.g., comparing sexual transmission like HIV with airborne transmission like SARS-CoV-2). Moreover, it influences the type of sample required and, consequently, the accessibility of the material
Incubation time	Extended incubation periods may permit serological diagnosis of acute infections and provide more time for responses (e.g., as seen in hepatitis). However, this could also delay containment efforts and result in a silent and broader spread of the disease
Pre- or asymptomatic spread	A clinical case definition can significantly raise the pretest probability compared to testing individuals who are presymptomatic or asymptomatic. Moreover, managing the spread of the disease will become considerably more challenging in case of significant pre-/asymptomatic transmission
Symptomatic spread	The duration of symptomatic spread and the detectability by diagnostic assays are critical factors
Postsymptomatic spread	The duration of spread after cessation of symptoms is relevant for models and disease control
Manifestations	Specific clinical manifestations can strengthen the effectiveness of a case definition, improve pretest probability, and increase the likelihood of isolation even in the absence of a positive test result. A high manifestation index, in turn, indicates a low rate of asymptomatic cases. In case of a zoonotic component, animals might also show disease
Severity/lethality	Ultimate outcome, higher severity of the disease, different perception especially of acceptable sensitivity for individual diagnosis
Therapeutic options	Timely diagnosis becomes crucial for the overall management of the outbreak if there are therapeutic options that yield favorable outcomes only when administered early. Furthermore, if the therapeutic option is very toxic or over-diagnosing can lead to severe side effects, especially in a low pretest probability setting, careful consideration is imperative
Preventive measures	The availability of vaccines, particularly those that provide personal and preferably sterilizing protection, significantly impacts the management of the outbreak. In addition, non-pharmaceutical measures such as hand hygiene, masking, and social distancing to prevent community infection or vector control are critical
Animal reservoir/vector	Most animals cannot be subject to normal infection control measures; thus, management of outbreaks with animal hosts has to include veterinary or vector control measures. Furthermore, animal material may be an essential resource for surveillance. Vector control measures may play a critical role in controlling the spread
Immunity	The presence of sterilizing immunity after infection allows stratification of subjects and potentially the development of herd immunity. Additionally, it serves as an indicator of the potential for developing an effective vaccine

HIV human immunodeficiency virus, SARS-CoV-2 severe acute respiratory syndrome coronavirus 2

Scenario Stratification: Socioeconomic Levels and Outbreak Phases

For developing impactful interventions, several levels of action should be taken into account. For example, the performance parameters required to achieve a significant impact may vary between public health-centered testing (such as testing on a population level) and individual testing, with implications for patient management. As a result, the diagnostic needs and TPPs for different scenarios should be categorized based on patient-related factors (such as pathogen detection and therapy monitoring). At the institutional level, where hospitals or clinics are inundated with suspect cases, diagnostics are necessary for severity prediction for triaging purposes. At the population level, diagnostics are needed to provide a safe environment for schools, workplaces, and mass gatherings, monitor transmission hotspots that require extra interventions, and for determining prevalence and immunity (Fig. 1).

Fig. 1.

Stratification of diagnostic needs and development of target product profiles. Dx diagnostics, TPP target product profile

In addition to the characteristics of the disease itself, discussions within the panel quickly revealed that the requirements for diagnostics vary greatly depending on the overall circumstances. This includes not only the socioeconomic context but also the phase in which the outbreak is unfolding. For example, in many LMICs, existing laboratory facilities are often critically inadequate, and public health systems, already constrained, are easily overwhelmed by the rapidly increasing patient load during an outbreak. To mitigate the pace of transmission through a population, many of these countries found it necessary to impose stringent lockdowns. However, in numerous LMICs, these lockdowns imposed severe and disproportionate economic hardship on populations, jeopardizing livelihoods and exacerbating poverty. Therefore, a diagnostic test that is affordable and widely available, even if it is not the most sensitive, becomes a crucial tool for providing safer environments and enabling economic activity to continue in settings where the economy is particularly vulnerable [7].

Therefore, it was necessary to define phases of an epidemic that focus on the rational use and impact of diagnostic tools, including their usefulness for non-therapeutic community mitigations. This simplified approach differs from other models that consider a wide range of factors, such as genetic adaptations or other events occurring during the emergence of a pandemic threat. Detailed phases, such as those defined by the World Health Organization for influenza [8], do not provide practical and actionable guidance for the diagnostic needs or the development of diagnostic products during different outbreak phases.

Three broad phases of infectious disease outbreaks that necessitate different diagnostic performance characteristics were defined. The unique medical needs arising at each pandemic phase are classified to ultimately guide the development of TPPs.

Phase 1—The Early Phase: In this phase, human cases have been reported and only a limited number of patients are known. The main effort is still to contain and understand the disease. Obviously, many characteristics of the disease are still unknown, and significant resources are being directed towards research to gain a better understanding of the pathogen, the pathophysiology, and disease progression. Highly sensitive and specific tests such as polymerase chain reaction (PCR), reverse-transcriptase PCR, or similar systems for pathogen detection and next generation sequencing, are needed to continue to refine and standardize the clinical case definition and to conduct research, including:

• The nature of the pathogen: what are its modes of transmission and in which sample material can sufficient concentrations of the pathogen be detected?

• What is the viral load threshold for transmission which should inform the development of the rapid antigen tests for phase 2?

• How many individuals are affected, who are the vulnerable populations, and are there asymptomatic or oligosymptomatic infections or underlying circulation in the past (underreporting factor)?

• Is there immunity against reinfection?

• Is there any animal or environmental host/vector/reservoir? Is the disease transmitted from animals to humans (where animals are the initial carriers and humans become infected from them), or do humans transmit the disease to animals?

As the design and manufacture of primers and probes are straightforward, as soon as genomic information on the pathogen is available, molecular tests can be quickly adapted. Additionally, it is anticipated that, in future outbreaks, sample material will contain sufficient nucleic acid for easy detection using these methods. If the pathogen is already well characterized, such as, for example, specific strains of influenza or coronaviruses, antigen-based rapid diagnostic tests (RDTs) with adequate sensitivity can be used to reliably quantify the number of cases detected through decentralized testing. Work on expedited specifier and RDT design is ongoing and continuously improving, but it is still not guaranteed that such a tool will be available in a timely manner for novel pathogens, as the development and production timelines for RDTs are longer compared to nucleic acid-based testing systems.

If serological tests (immunoglobulin M [IgM] or antigen) are available early on, they can play a valuable role in identifying acute infection. For example, IgM tests can be an important diagnostic tool in cases where molecular tests are not easily applicable due to limited accessibility of the pathogen (e.g., organ-bound or limited to cerebrospinal fluid). Also, molecular testing capacities might not be available, or waiting times for results of molecular tests are long. A serologic testing approach has been used in Peru and China with much success [9]. These assays may also provide evidence of possible immunity following infection in a small, well-characterized patient group. Thus, there are distinct diagnostic needs at both the individual and public health levels during this early phase of an outbreak (Table 2).

Table 2.

Important knowledge gaps and diagnostic needs in individual outbreak phases

1: Early phase	2: Waves/outbreak phase	3: Endemic phase
Focus on individual:
Reliably diagnose the disease in patients with certain symptoms Determine the mode of transmission (impact on behavior) Evaluate the material in which the pathogen is shed, along with the duration and pathogen concentration of the shedding Determine whether the disease is vector-borne (e.g., use repellants or other protective measures), or if it has another environmental reservoir or zoonotic potential Assess the presence of protective immunity	Identify/confirm individual cases (yes/no) Anticipate high-throughput testing as necessary in specific areas Provide prognosis for severe courses (triage, supplemental health care, need for high dependency/intensive care unit beds)	Identify/confirm individual cases (yes/no) Prepare for intermediate-throughput analyses Establish disease severity markers (triage) Implement monitoring routine on individual patient level as required
Focus on public health:
Characterize/identify the disease/pathogen Identify the mode of transmission and sample type Determine the number of affected subjects Determine the presence of silent infections/underlying circulation in the past (underreporting factor) Assess immunity against reinfection Identify animal host/vector if applicable Identify all cases and test contacts to potentially end a specific outbreak	Trace contacts Assess underreporting factor (epidemiology and model prediction) Perform genomic pathogen surveillance to detect immune escape and other variants and improve available diagnostic tools Perform vaccine and therapeutic efficiency evaluations if available and warranted	Perform epidemiological surveillance regarding genetics Understand epidemiology regarding immune status in the population to understand the further development of the disease (similar to influenza) Perform vaccine and therapeutic efficiency evaluations if available and warranted

Phase 2—The Waves/Outbreaks Phase: This phase is characterized by a significant increase in the number of patients, either globally or in massive numbers in localized single or multiple outbreak clusters. As a result of the absence of significant underlying population immunity, many subjects become infected in parallel. This often coincides with the overwhelming of healthcare facilities and diagnostic capabilities, at least in localized areas.

During this situation, the clinical case definition becomes crucial. The uniqueness of the case definition increases the pretest probability. In extreme situations, diagnostic tests may not provide significant benefits in identifying cases if the manifestation index is high and the clinical presentation is sufficiently specific. However, for diseases with mild or nonspecific symptoms, or a high percentage of asymptomatic cases, the need for diagnostic tools increases. This phase might also provide clarity on the extent to which asymptomatic cases influence the epidemiological dynamics. If the disease is fatal for many patients or leads to severe courses requiring hospitalization, the need for triage and emergency treatment centers arises. Test systems that can process a large number of samples and are adapted to potential new pathogen variants are essential.

Disease severity predictors can greatly alleviate the burden on the healthcare system. Home and decentralized testing are important to prevent the concentration of patients or sick individuals in a few locations, and can reduce further spread. This also helps alleviate the workload on overburdened laboratories. RDTs used on individuals matching the case definition may still provide considerable advantages even with lower sensitivity and specificity. If immunity is likely, epidemiological tools such as sero-surveys can provide important insights for authorities in terms of epidemiological data (e.g., extent of spread and population immunity), as well as guide future steps, such as containment measures or vaccine development. This phase is intended to study the relevance of the “immunological scar”, i.e., T cell responses as an indicator for the quality and longevity of immunity. However, such studies might further challenge the first-line diagnostic capacities, and collaboration of clinicians, academic partners, and diagnostic suppliers can help to balance shortages. The diagnostic needs are outlined in Table 2.

Key diagnostic foci:

• Identification/confirmation of individual cases (yes/no) (requiring high throughput in certain areas)

• Prognosis for severe courses (triage)

• Containment and contact tracing

• Estimation of the underreporting factor (epidemiology and model prediction)

• Test adaptation to new pathogen variants

• Investigate quality and longevity of putative immunity

Phase 3—The Endemic Phase: After one or multiple waves of infections, epidemic diseases generally either disappear completely or resurge later as outbreaks. In the latter case, this corresponds to phase 2, but with certain differences, such as increased underlying immunity within the population, if present. Alternatively, the disease may enter an endemic phase, as seen with SARS-CoV-2, characterized by a more or less permanent circulation of the disease, but with reduced intensity of infections. A certain level of normalization can be observed in the healthcare sector. Healthcare professionals become familiar with the clinical presentation, and the demand for diagnostics can be estimated, either year-round or seasonally. This allows for better resource allocation and alleviates the strain on the healthcare system caused by outbreaks. Over time, multiple diagnostic and therapeutic options become available, and there is a growing body of experience regarding their use, case definitions, and limitations. It is likely that diseases caused by such outbreak organisms will remain notifiable diseases for a while, necessitating confirmed diagnoses even in cases of mild manifestations. Even during the phase of permanent circulation, a certain level of surveillance will be required to avoid missing any recurrence of epidemiological threats. This surveillance may include monitoring case numbers, genetic surveillance to detect immune escape and other variants, as well as potentially monitoring immune correlates in the population.

Key diagnostic foci:

• Identification/confirmation of individual cases (yes/no) (requiring intermediate throughput, often more predictable)

• Potentially identifying disease severity markers (triage)

• Epidemiological surveillance to monitor genetic and immune status in the population in order to understand the further development of the disease (similar to influenza).

Target Product Profiles

It was discussed that the effectiveness of tests, taking into account their varying sensitivity, specificity, and access to sample materials, depends on factors such as the phase of a potential outbreak scenario, pretest probabilities, and available intervention strategies. Therefore, TPPs that outline assay requirements should take into account each outbreak scenario phase and weigh the importance of different parameters depending on the consequences.

Consensus regarding the parameters used to define TPPs for diagnostic test systems is currently lacking. Although sensitivity and specificity are commonly deployed as performance parameters, there is no agreement on the appropriate definition of the denominator or control group for accurately determining these parameters. Consequently, even if the sensitivity and specificity values appear identical on paper, the resulting predictive values and subsequent implications for patient management can differ significantly, making them difficult to interpret. Moreover, in outbreak situations, the desired TPP may vary considerably depending on the phase of the ongoing outbreak, as discussed earlier. Factors such as the number of patients, testing settings, accuracy of clinical case definitions, availability of vaccines or therapeutics, and other variables contribute to this variability.

Consequently, creating universal TPP specifications is not entirely feasible. Instead, it is more appropriate to identify the key parameters that should guide the development of a TPP. Subsequently, the TPP for each specific use case should be determined based on the practical applications of the test in real-world scenarios, while considering the potential consequences of the test results on treatment decisions, and broader political and public health considerations to help control the outbreak. It seems evident that a TPP can change with the onset of a pandemic scenario to perfectly match the needs of the respective phase. However, registration efforts and insecurity on pathogen evolution and epidemiological development favor a one-fits-all solution as far as resulting trade-offs are acceptable. The result is a succinct summary of these parameters, along with a brief rationale (Table 3).

Table 3.

Important parameters to be considered for target product profiles

Parameter	Rationale for why this is important
Target population	It needs to be defined who is at risk and who will be eligible for testing. The ability to access/obtain sample materials and their volume (pediatric, etc.), as well as the availability of a reliable clinical case definition, needs to be considered
Phase of outbreak	Diverse requirements beyond individual treatment decisions, e.g., in public health, assessing previous infections, evaluating infectiousness, pathogen’s genetic diversity, etc.
Target setting	The infrastructure varies significantly, e.g., home testing vs. health post vs. tertiary care center vs. reference laboratory; the ability to collect samples (sample materials) and the methods used (required laboratory machines, cold chain, training) vary dramatically depending on the setting
Sensitivity (%) and specificity (%)	Depends on pretest probability and is therefore defined for a specific target population. Needs may vary during different phases of an outbreak
Sample type	Determines the setting in which the diagnostic procedure can be performed
Sampling approach	Determines the setting in which the test can be performed as well as its accessibility. Sampling scenarios can vary from non-invasive home sampling by lay individuals to invasive sampling by clinical personnel in healthcare settings
Output result	An important parameter for test developers, which is either qualitative or quantitative. Quantitative assays are used when certain levels of the analyte—not just its presence—are of medical interest
Time to result to patient	Indicates the time from sampling until the patient is informed about the diagnosis
End user profile	Determines by whom and in which setting the test should be performed, influencing packaging, technology used, storage conditions, etc.
Description of diagnostic	Describes technical requirements and platform/technologies that can potentially be used

It is evident that the ideal diagnostic solution would be extremely rapid, performed at the bedside or at home by lay persons not requiring trained personnel, consistently accurate, and capable of using non-invasive samples such as breath. However, it is important to recognize that aiming for an unrealistic combination of parameters in a TPP is not practical. Instead, the focus should be on assessing the consequences of the test results and deriving specifications for an MVP that will still provide sufficient benefits. Any further improvements beyond the MVP would be desirable but not essential for the initial implementation of the system. The workshop identified the criteria for both the MVP and the specifications of a product that specialists in the field would find satisfactory, taking into account the psychological factors that influence treatment and testing decisions made by professionals.

Considering the multitude of variables that need to be taken into account, the experts were involved in defining specific parameters for disease case scenarios. On the basis of these scenarios, MVP-TPPs were developed for different use cases and phases of a polio-like virus outbreak scenario (Table 4).

Table 4.

Target product profile (TPP) showing desired and minimal viable product (MVP) requirements for diagnostic tests for infectious diseases at different stages of an outbreak (based on a polio-like virus outbreak scenario)

	Phase 1		Phase 2		Phase 3
	Desired	MVP	Desired	MVP	Desired	MVP
Diagnostic need	Identify individual patients for interventions		Identify transmitting individuals to guide public health interventions		Elucidate the percentage of the population still susceptible to infection and to identify vaccine breakthroughs
Target population	Symptomatic patients according to current case definition, people at risk, HCWs, and asymptomatic contacts of infected patients		Populations attending schools, mass events, hospitals, care homes, returning to work, etc		Representative general population
Target setting	Primary, secondary, ambulatory care centers		Community based		Seroprevalence studies, surveillance networks/sites
Sensitivity	Realistic 95%	80% (driven by R0)	90% (depending on threshold of transmission)	80% (driven by R0)	> 90%	80%
Specificity	> 98%	95%	97%	95%	95%	90%
Sample type	Minimally invasivea	Minimally invasivea	Simple non-invasive	Simple non-invasive	Blood/DBS	Blood/DBS
Sampling approach	Ambulatory	Not discussed	Convenience/any setting	Convenience/any setting	Ambulatory	Ambulatory
Output result	Qualitative	Qualitative	Qualitative	Qualitative	Not discussed	Not discussed
Time to convey result to patient	As fast as possible	< 24 h	15 min maximum	< 30 min	N/A	N/A
End user profile	HCWs	HCWs	Able to follow instructions	Able to follow instructions	Laboratory professional	Laboratory professional
Description of diagnostic test	Non-manual molecular	PCR-based (automated) or easy-to-use antigen or IgM test if molecular tests are not available	Fraud-proof, connective, easy to operate for lay persons	Fraud-proof, connective, easy to operate for lay persons	Serology (total Ab, IgG)	Serology (total Ab, IgG)

Ab antibody, DBS dried blood spot, h hours, HCWs healthcare workers, Ig immunoglobulin, min minutes, MVP minimal viable product, N/A not applicable, PCR polymerase chain reaction, R₀ basic reproduction number, TPP target product profile

^aSwab, capillary blood, feces, urine

Modeling

Furthermore, the impact of various testing scenarios in a polio-like outbreak scenario was modeled, with a focus on sensitivity, availability, and contact tracing. The model is based on a compartmental model of ordinary differential equations, described in more detail elsewhere [10]. In this model of infectious disease dynamics, individuals transition between mutually exclusive states (compartments) comprising susceptible (S), exposed (E), infected (I), detected (D), vaccinated (V), recovered (R), or dead (X). The model allows the inclusion of up to two disease stages for I and D (e.g., representing presymptomatic and symptomatic) and up to two vaccine stages (representing doses in a series). The numeric evaluation is based on an exemplary population of 80 million individuals, similar to the population of Germany. We investigated the impact of six different testing scenarios on the number of infected individuals in this model. These scenarios were no tests at all (base scenario); a scenario in which the test has 100% sensitivity, but only 1% of symptomatic individuals can be tested per day (scenario S1); a scenario in which the test has 50% sensitivity, but unlimited availability for symptomatic individuals (scenario S2); a scenario similar to S2, but with low effort tracing leading to 1% of pre- and asymptomatic individuals being tested daily (scenario S2low); a scenario similar to S2 but with higher (medium-level) effort tracing leading to 5% of pre- and asymptomatic individuals being tested daily (scenario S2medium); and a scenario similar to S2 but with high random testing efforts leading to 10% of pre- and asymptomatic individuals being tested daily (scenario S2high). The assumptions included in the model are summarized in Table 5.

Table 5 .

Assumptions in the model for a polio-like virus outbreak scenario

Causative agent	Polio-like virus (non-enveloped RNA virus)
Mode of transmission	Fecal–oral and aerosol Initially up to 30 days of viral shedding
R0	3–5
Incubation time	3–10 days
Presymptomatic spread	1–2 days
Symptomatic spread	15 days
Postsymptomatic spread	Up to 15 days
Quarantine measures	100% compliance with isolation upon positive test result
Manifestation	1–2%
Lethality	20% of symptomatic cases
Therapeutic options	None
Preventive measures	Not discussed
Animal reservoir	Unknown
Immunity	Sterilizing (permanent)

RNA ribonucleic acid

When comparing different test scenarios with either a 50%-sensitivity test with high availability (scenario S2), 100%-sensitivity test with very restricted availability (scenario S1), or no tests at all (base scenario) it could be shown that a perfect but slow test with limited availability has nearly no effect and that imperfect tests with high availability decreased the peak of infectious individuals in the population substantially by 8% (Fig. 2). As the manifestation index was very low, with 1.5%, testing symptomatic patients only had very limited effects. If a test with only 50% sensitivity was not expensive and available, testing pre- and asymptomatic patients could also be considered. This could either be achieved by completely random tests in the whole population or by targeting populations at risk via contact tracing. If 1% of the infected individuals were tested daily with 50% sensitivity (scenario S2), the peak decreased by 14%, doubling the effect of testing. If the tests were very easily available and the random tests could thus be increased to 10% of the infected population per day (scenario S2high), the curve flattened significantly and was right-shifted. The peak of infected individuals decreased by 70% compared to no testing. The peak of infectious individuals was right-shifted by 0 days in the case of the perfect but rarely available test (scenario S1) and shifted by 17 days with the highly accessible low-sensitivity test (scenario S2). The low, medium, or high effort tracing delayed the peak by 31 (scenario S2low), 107 (scenario S2medium), or even 285 (scenario S2high) days (Fig. 2).

Fig. 2.

Impact of different testing scenarios on the number of infectious individuals in the population. All tests take 1 day. The contact tracing does not necessarily have to be a formal tracing procedure by officials, but could also be a voluntary self-testing after having contact with infected individuals. All infected individuals are able to potentially spread the virus. The actual disease manifests in only 1.5% of them. Time in days. s0 scenario 0 (no tests at all); s1 scenario—100% sensitivity, only 1% of symptomatic individuals can be tested per day; s2—50% sensitivity, unlimited availability for symptomatic individuals; s2low –similar to s2 but with low effort tracing leading to 1% of pre- and asymptomatic individuals being tested daily; s2medium—similar to s2 but with higher effort tracing leading to 5% of pre- and asymptomatic individuals being tested daily; s2high—similar to s2 but with higher effort tracing leading to 10% of pre- and asymptomatic individuals being tested daily

Death rates were also decreased with increasing test availability, testing pre- and asymptomatic individuals, and contact tracing. The suboptimal testing scenario (scenario S2) already decreases deaths by 3%, which would have saved about 6700 lives in the scenario, while the highly sensitive but limited test (scenario S1) only decreases deaths by 0.05% (− 120 individuals) so the fast scenario saved more than 50 times more lives than the testing scenario with a perfect test and limited availability. Testing pre- and asymptomatic infected individuals further increased the effect. In the low tracing scenario (S2low), deaths were reduced by 5% (− 12,400 individuals), and considering the high tracing scenario (S2high), deaths were even reduced by 42% (− 99,000).

It is important to note that if test durations were longer than 1 day, the effect of the test strategies decreases dramatically (repeated modeling for 3, 5, and 7 days; data not shown).

Discussion

The current method of defining TPPs for MVPs relies largely on expert opinions. To standardize this process, an expert panel discussed and created a systematic approach to tailor TPPs specifically for different outbreak scenarios. This begins with identifying diagnostic needs and moves towards TPP development. A framework to define outbreak scenarios was provided, highlighting key characteristics. These scenarios were further categorized by various socioeconomic levels and phases to pinpoint diagnostic needs, driving the tailored development of TPPs. Additionally, crucial factors for TPP development were identified, and exemplary TPPs for different phases of a polio-like virus outbreak were formulated. Various polio-like virus testing scenarios, considering diagnostic tests with different sensitivities, availabilities, and contact tracing methods, were also modeled.

Currently, there is no unified consensus among regulators on the necessary specifications for diagnostic tests to gain licensing. Regulations such as the European In Vitro Diagnostic Regulation have been proposed, attracting significant criticism. For instance, during the early stages of the SARS-CoV-2 pandemic, there was a strong emphasis on achieving excellent test performance, which is certainly a desirable goal. However, factors other than high accuracy might play a crucial role in the overall impact of test results and consideration should be made to provide reference samples to calibrate the assay and thus ensure performance integrity. In the USA, despite the availability of highly sensitive RT-PCR tests, the initial number of identified cases was very low. Consequently, even with a sensitivity greater than 90%, less than 10% of cases in the entire population were diagnosed due to the limited number of tests conducted, caused by issues of availability and capacity [11].

A test with only 50% sensitivity might be dismissed as unreliable, yet in such circumstances, it could potentially catch more cases than a high-sensitivity test with limited availability. For example, in the SARS-CoV-2 scenario, using a home test with 50% sensitivity on 1000 random individuals would detect five times more infected people than using a test with 100% sensitivity on just 100 individuals. For public health measures and control, such a lower-sensitivity test could provide more actionable insights and a better overview of the situation than a higher-quality but less accessible test. This was supported by modeling different testing strategies in a polio-like virus outbreak.

This approach of focusing on test availability, rather than sensitivity alone, is more effective in a pandemic situation. Given that often a large share of infected individuals are asymptomatic, random testing of the general population or through contact tracing is crucial. Making tests highly affordable and widely available and accessible could encourage more individuals to undergo testing voluntarily or comply with public regulations, aiding in the detection of asymptomatic cases. This was evident in parts of Africa including Cameroon, where widespread availability of rapid testing in public spaces such as marketplaces allowed the government to ease strict measures, thereby reducing the economic impact on the country [12, 13]. This knowledge could help decision-makers in defining the specifications for diagnostic tests in outbreak situations.

The question of funding for diagnostic developments must be raised. While clear business cases are being developed by diagnostics manufacturers, the cost for such developments is prohibitive for rare diseases or diseases encountered in low-income settings. Also, diagnostic solutions for diseases suspected to cause outbreaks in the future, but unlikely to allow immediate market access, will not be developed by for-profit companies. Thus, public–private partnerships or publicly funded development or research initiatives will have to step up in such cases to improve preparedness and reaction time.

Lastly, we must acknowledge limitations and potential sources of bias in our model and report. Some factors that may influence infection numbers were not incorporated in our model, including the impact of preventive measures (such as isolation), compliance with such measures, and immunity waning. To keep the model simple, we deliberately chose a virus that is spread by human-to-human contact; therefore, the findings may not be applicable to vector-borne diseases. We also assumed 100% compliance with isolation measures among individuals testing positive, which is probably optimistic. Nevertheless, it is difficult to see how inclusion of vector-borne dynamics or poor compliance with quarantine recommendations would affect the overall conclusions regarding TPP characteristics of diagnostic tests by outbreak phase. In addition, compartmental models inherently simplify infection dynamics within a population, and cannot capture all of the potential influences on infection rates, including individuals’ innate resistance to disease or immune response. While every attempt was made to be unbiased in our assessment of TPPs and to develop an “infection-agnostic” model, we cannot rule out the possibility of unconscious bias influencing our process and conclusions.

Conclusions

This expert panel-led investigation and subsequent modeling of different testing strategies sought to quantify the impact of trade-offs between accessibility, affordability, and speed of diagnosis in different outbreak settings. On the basis of the findings presented above, diagnostic tests used during an outbreak should prioritize speed, cost-effectiveness, and broad availability—including ease of access to the diagnostics market—even if it means compromising on sensitivity.

Author Contributions

All members of the expert panel are authors of this paper. All authors meet the International Committee of Medical Journal Editors (ICMJE) authorship guidelines for this manuscript. Conceptualization: Markus Beutler, Matthias Strobl, Andreas Wieser. Methodology: Markus Beutler, Christoph Sticha-Kaiser, Noemi Castelletti, Matthias Strobl, Andreas Wieser. Software: Christoph Sticha-Kaiser, Noemi Castelletti. Investigation: Markus Beutler, Pablo Bonvehí, May Chu, Veasna Duong, Rosanna W. Peeling, Jonas Schmidt-Chanasit, Simon Jochum, Christoph Sticha-Kaiser, Noemi Castelletti, Matthias Strobl, Andreas Wieser. Data curation: Christoph Sticha-Kaiser, Noemi Castelletti. Manuscript drafting: Markus Beutler, Andreas Wieser. Manuscript review and editing: Markus Beutler, Pablo Bonvehí, May Chu, Veasna Duong, Rosanna W. Peeling, Jonas Schmidt-Chanasit, Simon Jochum, Christoph Sticha-Kaiser, Noemi Castelletti, Matthias Strobl, Andreas Wieser. Data visualization: Christoph Sticha-Kaiser, Noemi Castelletti. Project administration: Markus Beutler. Supervision: Markus Beutler, Matthias Strobl, Andreas Wieser. Acquisition of funding: Markus Beutler, Simon Jochum, Matthias Strobl.

Funding

BayVFP research grant on developing diagnostic solutions for parameters with pandemic potential (LSB-2103-0002), and funding from Roche Diagnostics International and Roche Diagnostics GmbH. The journal’s Rapid Service Fee was funded by Roche Diagnostics GmbH.

Data Availability

Requests concerning the data that support the findings of this study can be directed to rotkreuz.datasharingrequests@roche.com for consideration.

Declarations

Conflict of Interest

Markus Beutler is an employee of Roche Diagnostics GmbH. Simon Jochum and Matthias Strobl are employees of Roche Diagnostics GmbH and hold shares in F Hoffman-La Roche. Pablo Bonvehí, May Chu, Veasna Duong, Rosanna W. Peeling, Jonas Schmidt-Chanasit, and Andreas Wieser received an honorarium from Roche Diagnostics for their participation in the workshop, and funding for related travel expenses, hotel costs, and meals; no payment was made for authorship. Andreas Wieser has received honoraria and advisory board fees from Roche Diagnostics in the past 5 years and has received discounted rates for some Roche Diagnostics products. Christoph Sticha-Kaiser is an employee of the Fraunhofer Institute for Translational Medicine and Pharmacology and has a collaboration contract with the Institute of Infectious Diseases and Tropical Medicine, LMU University Hospital. Noemi Castelletti has no conflicts of interest to report.

Ethical Approval

This study involved a model based on the discussions of an expert panel, and did not use real patients or samples. Ethical approval was, therefore, not required.