To establish strategic priorities for the German national public health institute (RKI) and guide the institute's mid-term strategic decisions, we prioritized infectious pathogens in accordance with their importance for national surveillance and epidemiological research.
We used the Delphi process with internal (RKI) and external experts and a metric-consensus approach to score pathogens according to ten three-tiered criteria. Additional experts were invited to weight each criterion, leading to the calculation of a median weight by which each score was multiplied. We ranked the pathogens according to the total weighted score and divided them into four priority groups.
127 pathogens were scored. Eighty-six experts participated in the weighting; “Case fatality rate” was rated as the most important criterion. Twenty-six pathogens were ranked in the highest priority group; among those were pathogens with internationally recognised importance (e.g., Human Immunodeficiency Virus, Mycobacterium tuberculosis, Influenza virus, Hepatitis C virus, Neisseria meningitides), pathogens frequently causing large outbreaks (e.g., Campylobacter spp.), and nosocomial pathogens associated with antimicrobial resistance. Other pathogens in the highest priority group included Helicobacter pylori, Respiratory Syncytial Virus, Varicella zoster virus and Hantavirus.
While several pathogens from the highest priority group already have a high profile in national and international health policy documents, high scores for other pathogens (e.g., Helicobacter pylori, Respiratory syncytial virus or Hantavirus) indicate a possible under-recognised importance within the current German public health framework. A process to strengthen respective surveillance systems and research has been started. The prioritization methodology has worked well; its modular structure makes it potentially useful for other settings.
Citation: Balabanova Y, Gilsdorf A, Buda S, Burger R, Eckmanns T, Gärtner B, et al. (2011) Communicable Diseases Prioritized for Surveillance and Epidemiological Research: Results of a Standardized Prioritization Procedure in Germany, 2011. PLoS ONE 6(10): e25691. doi:10.1371/journal.pone.0025691
Editor: Igor Mokrousov, St. Petersburg Pasteur Institute, Russian Federation
Received: July 5, 2011; Accepted: September 8, 2011; Published: October 4, 2011
Copyright: © 2011 Balabanova et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study was funded internally by Robert Koch Institute. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The large number of infectious agents with different pathogen-specific, host-specific and socio-economic characteristics makes the allocation of the limited resources available within the area of prevention and control of communicable diseases both challenging and controversial. The amount of attention, efforts and funds spent on surveillance, control and research of infectious pathogens varies greatly between pathogens, settings and over time. This distribution often appears to be ambiguous, potentially guided by senior leaders' research interests, short-term political agenda or residuals of historic situations . As many pathogens are potentially harmful for humans and may present serious public health threats, it is necessary to prioritize the resources dedicated for surveillance and epidemiological research of infectious diseases. This needs to be done sensibly and rationally bearing in mind different aspects of pathogens' characteristics, their impact on societies and long-term consequences of their presence or introduction into populations.
The rational and transparent setting of priorities for investment into health research is therefore becoming an essential part of research planning –. Usefulness of prioritization, irrespective of its methodology, has been demonstrated by several research groups –. Prioritization can provide directions for future resource allocation and strategic planning at different levels (institutional, regional, national or international) and act as a platform for inter-disciplinary debate involving decision-makers, researchers, clinicians and the general public , .
Although today there are a number of published tools to guide the process of setting priorities, only a few publications openly describe the methodology in sufficient detail and transparency to allow reproducibility or adaptation in other settings , , , . Furthermore, very little is published in terms of actual prioritization results.
The Department for Infectious Disease Epidemiology of the Robert Koch Institute (RKI), German national public health institute, is in charge of national surveillance, prevention, control strategies and epidemiological research in the field of infectious diseases. Together with external senior experts the Department initiated a prioritization process aiming to (1) develop a rational system for setting priorities in the area of infectious diseases using a metric-consensus approach, and (2) rank most common pathogens in accordance with their importance for national surveillance and epidemiological research to guide future work of the RKI.
In the absence of established standards we designed a methodology using elements of our previous work in 2004 ,  and experiences of other groups , , , , . Our multi-staged prioritization process included compilation of the list of pathogens to be prioritized, development of evaluation criteria, weighting of criteria and scoring of the pathogens. The methodology was based on the core domain for good practices in setting priorities for research in health, such as legitimacy and fairness , , and represents the further development of the work from 2004.
Delphi process participants
The core team (YB, AG, GK) contacted relevant leading national public health institutions with the request to name a representative to take part in the Delphi consensus process that aimed at assessing the feasibility of the methodology and the relevance of the suggested criteria, as well as to discuss possible improvements and pathogens' scores. Ten external senior experts (BG, UG, JH, TJ, MK, MKr, TL, MP, NS, UU) were nominated. They represented the National Committee of Infectious Disease Epidemiology, the German Society of Hygiene and Microbiology, the German Society of Infectious Disease Specialists, the German Society of Epidemiologists, the National Reference Laboratories, the German Federal Ministry of Health, the RKI Scientific Committee, the German Federal Chamber of Physicians and the German Medical Association. Ten internal experts (GK, AG, YB, SB, RB, TE, OH, KS, OW, MM) represented units and departments of the RKI working in the field of infectious diseases. The participants shared expertise in bacteriology, virology, mycology, parasitology, general infectious diseases, tropical medicine, general medicine, epidemiology, public health, veterinary health and infection control.
Selection of pathogens
To maintain a broader approach, we decided to evaluate pathogens rather than diseases for their importance. A list of pathogens was compiled according to the following selection criteria: (a) notifiable according to the German law for the control of infectious diseases , (b) reportable within the European Union , (c) reportable to the WHO under the International Health Regulations , (d) agents with potential for deliberate release , and (e) pathogens represented in dedicated chapters in an established infectious diseases manual  and occurring in Germany. Some pathogens were grouped together when biologically and clinically plausible. The list of pathogens was reviewed by the RKI experts and the Delphi process participants.
Prioritization criteria and scoring
The twelve criteria used during our previous prioritization process in 2004 ,  were further modified according to the feedback received from a broad group of different experts . The newly suggested criteria and their three-tiered definitions were then reviewed by internal and external experts. The initial scoring of all pathogens according to each criterion was performed by the core team and internal experts from the RKI. The data supporting the scoring decisions and references to the data sources, or experts' own explanations were recorded in a structured format. The internal scoring was followed by a modified two-round Delphi consensus process (internal round with RKI and joint round with additional external experts) where scores were discussed.
Independent of the scoring, we invited a panel of external experts to assign a weight to each criterion. This invitation was sent to all 16 federal public health institutions, all 18 national reference centers, 49 consulting laboratories, 9 scientific and professional societies working in the area of control, prevention and research of infectious diseases and to all 72 participants of the 2006 online survey . External experts were asked to assign a value ranging from 0 to 10 to each criterion, thus reflecting the criterion's contextual importance for surveillance and epidemiological research. The value 0 reflects the lowest and 10 the highest level of importance of a criterion. More than one criterion could be assigned the same weight, similarly to techniques used in other prioritization exercises . The final criterion's weight was defined by the median value of all weights assigned by the participating experts.
Ranking of the pathogens
Each score was multiplied by the weight for the respective criterion. The sum of these weighted scores reflects the total weighted score of a pathogen. The total weighted scores were finally re-scaled to a range from 0 to 100 in order to facilitate final interpretation.
Following the experience from Canada , we did not focus on the exact numerical score assigned to a pathogen. The interpretation of the final weighted scores and their corresponding sequential ranks was done in priority groups reflecting four priority levels (the highest, high, medium and low priority). The cut-off limits for the groups were based on the equal ranges of 25 (0–25, 26–50, 51–75, 76–100). Distribution of the pathogens was later compared with their positions in the 2004 priority list.
Selection of pathogens and scoring
In total 127 pathogens were selected for prioritization. Drug-resistant strains were scored under the common pathogen group and were not assessed as a separate entity. During the Delphi rounds, changes in pathogens' scores were made based on the consensus approach.
The Delphi process also resulted in the recommendation to remove the criterion “Emerging potential” due to its ambiguous meaning (i.e. each infectious pathogen has an ability to emerge or re-emerge) and to rephrase the definitions of some criteria. The final criteria and their definitions are presented in Table 1. The detailed scores are available in the Table S1.
All criteria were weighted by a total of 86 experts (14 RKI and 72 external experts). All Delphi discussion participants took part in the weighting.
The opinions regarding importance of individual criteria varied greatly across the participants for some criteria, while it was similar for others. For example, eleven participants considered the criterion “Public attention” to be of a low importance (scoring it “2”) and the same number considered it to be of a relatively high importance (scoring it “6”). At the same time, other criteria such as “Case fatality rate (CFR)” and “Trend” were weighted in a similar way by the majority of the respondents. When looking at the median weights, the criterion “Case fatality rate” was considered the most important criterion (median weight of “9”) while “Trend” and “Public attention” (both with a median weight of “5”) were considered the least important ones (Table 2).
We analyzed weights assigned by the experts working in different areas of medicine: epidemiologists and public health specialists (n = 43), laboratory specialists (n = 35) and clinicians (n = 8) (Table 2). Several criteria were assessed similarly across the groups (for example, the criteria “Prevention”, median weight of “8”, or “Trend”, median weight of “5”). At the same time, the criterion “Incidence” was seen as one of the most important by epidemiologists and public health experts (median weight of “8”) and laboratory specialists (median weight of “7”) but was seen as one of the least important by clinicians (median weight of “5.5”). Almost the reverse situation was observed with the criteria “Work and school absenteeism” and “Health care utilization” that were seen to be important by clinicians but perceived as of lower importance by the two other groups.
Table 3 presents the allocation of the pathogens into four priority groups according to their weighted total score. The highest priority group contains 26 (20.5%) pathogens. Among those are the pathogens that already received the highest priority in our previous prioritization process, such as HIV, Mycobacterium tuberculosis, Staphylococcus aureus including methicillin resistant strains (MRSA), Influenza virus, Hepatitis C virus, Campylobacter spp., Neisseria meningitidis, Legionella pneumophila, Varicella zoster virus (VZV). It additionally contains a number of pathogens responsible for nosocomial infections (e.g., Klebsiella spp., Pseudomonas ssp., Enterococcus spp.) and Respiratory syncytial virus (RSV) that were not evaluated by us previously. Helicobacter pylori and Hantavirus now belong to this group, while in 2004 both pathogens held medium priority positions. Other pathogens that were scored highest in 2004 were now assessed to be in the medium priority group (e.g., Parvovirus B19).
The high priority group contains 39 (30.7%) and the medium priority group - 45 (35.4%) pathogens. A number of pathogens that were ranked as the least important in 2004 (Vibrio cholera, Francisella tularensis, Bacillus anthracis, Bartonella quintana) were now assigned to the medium and high priority groups.
Out of 17 pathogens from the low priority group in 2011, 11 were newly added.
This prioritization approach allowed us to benefit from the cumulative knowledge of many experts. The results of our work seem convincing to us; they support our current activities as well as indicate new directions for the future work. For example, the ranking of the majority of the pathogens found in the highest priority group (e.g., HIV, M.tuberculosis, Influenza virus, Neisseria meningitides, Legionella pneumophila) is in line with strategic goals identified by a number of international agencies that focus both on resource-constrained health systems and on industrialized countries such as Germany , , –. The decision to evaluate a broad range of nosocomial pathogens in the current prioritization and their high ranks indicate a growing recognition of the problem of antimicrobial resistance and healthcare associated infections (HAI) and are in line with a number of new strategic international and national policies calling for enhancement of HAI surveillance activities and capacities , .
The positioning of Helicobacter pylori, Hantaviruses, RSV and VZV among the pathogens in the highest priority group helped us to identify the under-recognized importance of the diseases with respect to surveillance and epidemiological research and call for actions in this respect.
Indeed, although there is a large amount of clinical and laboratory research dedicated to Helicobacter pylori and a number of clinical guidelines are available, very little is done in terms of public health surveillance, despite the growing rates of antimicrobial resistance seriously compromising treatment , .
RSV remains the most common respiratory pathogen in infants and young children worldwide, often resulting in serious lower respiratory tract infections , yet a routine surveillance system for RSV based on virological testing of sentinel respiratory samples has only been initiated recently, and the burden of disease at the population level in Germany is largely unknown. The placement of RSV in the highest priority group is particularly unusual as pathogens causing diseases in limited population groups, i.e. here in young children, are often believed to be of a lower overall importance.
The incidence of human disease caused by Hantavirus fluctuates significantly over time; its incidence has peaked in endemic areas in Germany in recent years, which may have contributed to the higher ranking of this pathogen in the 2011 prioritization and the call to enhance research activities. A population-based seroprevalence survey will indeed be initiated within the RKI-funded network of reference laboratories.
VZV, a virus that causes two frequent diseases in children and adults, was ranked as a pathogen with the highest importance both in 2004 and in 2011. Country-wide sentinel surveillance has been recently implemented for VZV after the implementation of a routine childhood Varicella vaccination program in 2004 .
The low priority group contains both pathogens with very low incidence in Germany (e.g., Mycobacterium leprae or helminths) and those much more common (e.g., Roseolovirus or Chlamydia pneumoniae), supporting our approach that the importance of a pathogen is defined by multiple factors rather than by its incidence or prevalence alone. Complex surveillance systems exist for some pathogens that were assigned to the medium priority group, for example, Francisella tularensis or Yersinia pestis. Although the amount of research that should be dedicated to those medium- and low-priority pathogens must be revised, undoubtedly the need to investigate outbreaks associated with these pathogens, to maintain diagnostic capacity and to continue efficient surveillance to pick up increasing trends of these rare illnesses, remains.
Similarly to other research groups that initiated processes for setting priorities , , , we found the compilation of the data necessary for the scoring process to be challenging due to a limited availability of reliable and published information. Participants did not find it easy to maintain sufficiently broad judgements despite the highly specialized expertise and focused research interest of some experts. Some degree of subjectivity can therefore never be avoided. Another limitation of our study was the difficulty to appropriately account for the decreasing trends of vaccine-preventable diseases that followed successful implementation of effective prevention programmes. Therefore, relatively low scoring of the respective pathogens for some of the criteria should not question the need to maintain adequate surveillance capacity for these illnesses.
In the applied methodology, we aimed at following the main principles of good practice in setting priorities in health research  and to reach maximum levels of objectivity, transparency, and reproducibility by integrating the following components: 1) a broad but systematic and reproducible selection of pathogens, rather than respective diseases to be included in the prioritization, 2) an explicit definition of each of the ten criteria using a three-tiered grading approach, 3) a comprehensive individual pathogen- specific scoring according to the best available evidence reviewed by a multidisciplinary expert group, and 4) a separate weighting of the criteria based on the involvement of a broad range of internal and external experts. Although this approach required intensive preparation, we believe it assured a high level of objectivity and transparency as demanded by Nuyens et al .
Our pathogen-specific approach allowed us to conduct prioritization not influenced by programmatic views and to compare pathogens' ranking within a group of diseases as well as across the groups (e.g., a pathogen being targeted by an antimicrobial resistance programme as well as by a zoonoses programme). This approach helped us uncover some differences in the importance of pathogens belonging to the same group, e.g., Neisseria gonorrhoeae and Trichomonias vaginalis.
Marked differences were observed in the weighting of the criteria between different professional groups. As the invitation to participate in the weighting was sent to various institutions and often forwarded further, it is not feasible to estimate the response rate. However, we received responses from at least one member of each contacted institution. The weighting of criteria is likely to correlate with societal values and reflect socio-economic, cultural and health system structure in a country. The fact that in Germany the criterion “Case fatality rate” was considered to be of the most importance, may reflect the high level of individualization and relative affluence of the German health care system. Comparisons to other countries are difficult since similar studies are lacking. One of the few other priority-setting initiatives with an explicit weighting procedure was conducted in Spain in which the criterion “Burden and importance of illness” was identified to be among the first three most relevant from a total of nine criteria, the other two being “Potential to change health outcomes” and “Potential to translate new knowledge into clinical or health services practices” . We can conclude that weighting needs to be explicitly addressed in any prioritization exercise and it is particularly appropriate for a national public health institute in order to reflect expectations from a broad range of professionals as the respective stakeholders . One way to even expand the societal perspective of the prioritization exercise could be to involve patients' representatives, similarly to how it has been done by Gooberman-Hill et al , for example.
Conclusions: The prioritization methodology presented here is based on the systematic evaluation of evidence and the involvement of a broad range of external experts. We feel that the results provide internal consistency and are plausible in the public health perspective. Our comprehensive and transparent approach makes the results defensible and shall give guidance for current needs in surveillance and epidemiological research in Germany. The list of ranked pathogens established here will serve as a reference for our mid-term strategic decisions, which will include strengthening the existing or introduction of new surveillance systems for pathogens from the high priority group (e.g., RSV, VZV or Helicobacterpylori) and re-consideration of the research and surveillance needs for those from the lowest priority group. It has already influenced the decision process on the need for the installation of new and continuation of existing national reference centers in Germany and the internal planning for the respective allocation of resources (GK personal communication). We plan to conduct a re-assessment of priorities within a five-year time frame based on the same methodology. The prioritization tool or its components can be applied across different areas of infectious diseases (by re-weighting prioritization criteria by different professional groups for different purposes) and in different geographical areas (by re-scoring pathogens according to their characteristics relevant for particular countries or continents). We hope that the presentation of our methodology could be helpful to other institutions that choose to prioritize their resources based on a transparent and standardized process.
Communicable diseases prioritized for surveillance and epidemiological research. Results of pathogen scoring.
We would like to thank all members of the Department for Infectious Disease Epidemiology, RKI for useful discussions, expert consultations and for taking part in the pathogens' scoring: C. Adlhoch, M.Altmann, K.Alpers, M.Askar, K.Atzpodien, B. Bartmeyer, H.Bernard, J.Benzler, J. Bielecke, B.Brodhun, U.Buchholz, W.Cai, S.Dudareva, M.Faber, G.Falkenhorst, L.Fiebig, C.Frank, B.Greutélaers, M.Dehnert, W.Haas, B.Hauer, W.Hellenbrand, J.Hermes, M.Kang, J.Koch, K.Köpke, S.Lassen, G.Laude, U.Marcus, D.Matysiak-Klose, A.Milde-Busch, M.Nachtnebel, S.Nielsen, I.Noll, R.Offergeld, S.Reiter, C.Remschmidt, A.Reuß, U.Rexroth, T.Rieck, B.Rosner, I.Schöneberg, B.Schweickert, A.Siedler, P. Stöcker, T.Süß, A.Takla, B.Ultsch, E.Velasco, L.Voß, M.Wadl, B.Weiss, D.Werber, H.Wilking, and R.Zimmermann.
We thank as well all colleagues who took part in the weighting procedure and Blair Prochnow for editorial support in the finalization of the manuscript.
Conceived and designed the experiments: YB AG GK. Performed the experiments: YB AG GK SB TE WH OH KS OW RB BG UG JH TJ M.Kist M.Kramer MM TL MP NS UU. Analyzed the data: YB AG GK SB TE WH OH KS OW RB BG UG JH TJ M.Kist M.Kramer MM TL MP NS UU. Contributed reagents/materials/analysis tools: YB AG GK. Wrote the paper: YB AG GK.
- 1. Daniels N (2000) Accountability for reasonableness. BMJ 321: 1300–1301.
- 2. Sibbald SL, Singer PA, Upshur R, Martin DK (2009) Priority setting: what constitutes success? A conceptual framework for successful priority setting. BMC Health Serv Res 9: 43.
- 3. Viergever RF, Olifson S, Ghaffar A, Terry RF (2010) A checklist for health research priority setting: nine common themes of good practice. Health Res Policy Syst 8: 36.
- 4. European Center for Diseases Prevention and Control (ECDC) ECDC Annual Work Programme 2011. Available: http://wwwecdceuropaeu/en/aboutus/Key%20Documents/101017_KD_ECDC_Annual_Work_Programme_2011pdf. Accessed 2011 April 12.
- 5. Berra S, Sanchez E, Pons JM, Tebe C, Alonso J, et al. (2010) Setting priorities in clinical and health services research: properties of an adapted and updated method. Int J Technol Assess Health Care 26: 217–224.
- 6. Fleurence RL, Torgerson DJ (2004) Setting priorities for research. Health Policy 69: 1–10.
- 7. Nuyens Y (2007) Setting priorities for health research: lessons from low- and middle-income countries. Bull World Health Organ 85: 319–321.
- 8. Priority Setting Methodologies in Health Research: A workshop convened by WHO's Cluster on Information EaRI, its Department for Research Policy and Cooperation (RPC) and the Special Programme for Research and Training in Tropical Diseases (TDR) (2008) Available: http://apps.who.int/tdr/stewardship/pdf/Priority_setting_Workshop_Summary10_04_08.pdf. Accessed 2011 May 15.
- 9. Remme JH, Blas E, Chitsulo L, Desjeux PM, Engers HD, et al. (2002) Strategic emphases for tropical diseases research: a TDR perspective. Trends Microbiol 10: 435–440.
- 10. Doherty JA (2000) Establishing priorities for national communicable disease surveillance. Can J Infect Dis 11: 21–24.
- 11. Carter AO (1992) Setting priorities: the Canadian experience in communicable disease surveillance. MMWR Morb Mortal Wkly Rep 41: Suppl79–84.
- 12. Lomas J, Fulop N, Gagnon D, Allen P (2003) On being a good listener: setting priorities for applied health services research. Milbank Q 81: 363–388.
- 13. Ghaffar A (2009) Setting research priorities by applying the combined approach matrix. Indian J Med Res 129: 368–375.
- 14. Krause G (2008) Prioritisation of infectious diseases in public health–call for comments. Euro Surveill 13:
- 15. Krause G (2008) How can infectious diseases be prioritized in public health? A standardized prioritization scheme for discussion. EMBO Rep 9: Suppl 1S22–27.
- 16. Gesetz zur Verhütung und Bekämpfung von Infektionskrankheiten beim Menschen (Infektionsschutzgesetz -IfSG) (2000) Bundesgezetz vom 20.Juli 2000. Bundesgezetzblatt 2000; 33(Teil I-G5702): 1045–77.
- 17. The Commission of the European Communities (2009) Commission Decision of 2 April 2009 amending Decision 2000/96/EC as regards dedicated surveillance networks for communicable diseases (notified under document number 2009/312/EC). Official Journal of the Europeam Union.
- 18. World Health Organisation (2008) International Health Regulations (2005). 2nd edition. Geneva, Switzerland: World Health Organization Press.
- 19. Tegnell A, Van Loock F, Baka A, Wallyn S, Hendriks J, et al. (2006) Development of a matrix to evaluate the threat of biological agents used for bioterrorism. Cell Mol Life Sci 63: 2223–2228.
- 20. Heymann D (2008) Control of communicable diseases manual. Washington: American Public Health Association.
- 21. Gilsdorf A, Krause G (2011) Prioritisation of infectious diseases in public health: feedback on the prioritisation methodology, 15 July 2008 to 15 January 2009. Euro Surveill 16:
- 22. World Health Organisation (2010) Global tuberculosis control 2010. Geneva: WHO/HTM/TB/20107.
- 23. United Nations (2010) Millenium Development Goals Report. MDG Report 2010.
- 24. Sixty-third World Health Assembly (2010) Medium-term strategic plan 2008–2013 (ammended). Availabe: http://appswhoint/gb/e/e_amtsp3html. Accessed: 2011 May 24.
- 25. German Federal Ministry of Health in association with German Federal Ministry of Consumer Protection FaAaGFMoEaR (2008) German Strategy Against Antibacterial Resistance [Deutsche Antibiotika-Resistenzstrategie, DART]. Berlin.
- 26. Tan VP, Wong BC (2011) Helicobacter pylori and gastritis: Untangling a complex relationship 27 years on. J Gastroenterol Hepatol 26: Suppl 142–45.
- 27. Wueppenhorst N, Stueger HP, Kist M, Glocker E (2009) Identification and molecular characterization of triple- and quadruple-resistant Helicobacter pylori clinical isolates in Germany. J Antimicrob Chemother 63: 648–653.
- 28. Miller S (2010) A community health concern: respiratory syncytial virus and children. J Pediatr Nurs 25: 551–554.
- 29. Siedler A, Arndt U (2010) Impact of the routine varicella vaccination programme on varicella epidemiology in Germany. Euro Surveill 15:
- 30. Kapiriri L, Tomlinson M, Chopra M, El Arifeen S, Black RE, et al. (2007) Setting priorities in global child health research investments: addressing values of stakeholders. Croat Med J 48: 618–627.
- 31. Gooberman-Hill R, Horwood J, Calnan M (2008) Citizens' juries in planning research priorities: process, engagement and outcome. Health Expect 11: 272–281.