gms | German Medical Science

GMDS 2015: 60. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

06.09. - 09.09.2015, Krefeld

No evidence of two distinct types of selection bias in physician-mediated long-term follow-up of malignant melanoma patients

Meeting Abstract

  • Sylke Ruth Zeissig - Institut für Med. Biometrie, Epidemiologie und Informatik (IMBEI), Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Veronika Weyer - Institut für Med. Biometrie, Epidemiologie und Informatik (IMBEI), Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Katharina Emrich - Institut für Med. Biometrie, Epidemiologie und Informatik (IMBEI), Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Sabine Fischbeck - Klinik für Psychosomatische Medizin und Psychotherapie, Medizinsche Psychologie und Medizinsche Soziologie, Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Barbara H. Imruck - Klinik für Psychosomatische Medizin und Psychotherapie, Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Peter Friedrich-Mai - Klinik für Psychosomatische Medizin und Psychotherapie, Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Manfred E. Beutel - Klinik für Psychosomatische Medizin und Psychotherapie, Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Maria Blettner - Institut für Med. Biometrie, Epidemiologie und Informatik (IMBEI), Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland
  • Harald Binder - Institut für Med. Biometrie, Epidemiologie und Informatik (IMBEI), Universitätsmedizin der Johannes Gutenberg-Universität Mainz, Deutschland

GMDS 2015. 60. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Krefeld, 06.-09.09.2015. Düsseldorf: German Medical Science GMS Publishing House; 2015. DocAbstr. 133

doi: 10.3205/15gmds165, urn:nbn:de:0183-15gmds1654

Published: August 27, 2015

© 2015 Zeissig et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at http://creativecommons.org/licenses/by/4.0/.


Outline

Text

Introduction: Register based studies have the advantage of access to a representative sample of patients in a certain catchment area. Therefore a cross sectional study (MeLa-study) of long-term melanoma patients recruited study participants in cooperation with the Cancer Registry Rhineland-Palatinate. In Rhineland-Palatinate direct access to patients via the cancer registry is prohibited by law for data protection reasons. Therefore, physicians who had reported their patients to the cancer registry had to be located and motivated to send letters to their former patients, more generally providing information on advantages and disadvantages of such an indirect access. We describe recruitment procedures and results in detail, and analyze whether responders and non-responders differ in terms of personal characteristics and course of disease. Differences between physicians (hospital or office-based; working in a rural or an urban setting), who were asked to take part in the study, and the corresponding patient groups are described. We examine if there are different effects on the participation proportion or data quality between patients, who were contacted by hospitals or those contacted by practitioners. Advantages and disadvantages of this indirect contacting of study participants are discussed and recommendations for further studies with similar recruitment procedures are presented.

Methods: Recruitment proportions for different groups of dermatologists (hospital vs. medical practice) are calculated. Response proportions of patients are assessed according to the method presented by Slattery et al. [1] with slight modifications (regarding the two-step approach of recruitment). Chi-Square and Fisher’s Exact Test are used to detect differences between participants and non-participants. Logistic regression models are used to assess the independent effect of
a) various patient characteristics on study enrollment regarding participation of study subjects and early-/late-participation respectively,
b) various characteristics of dermatologists on participation of the physician (including number of contacts leading to participation) as well as participation rate of contacted patients,
c) various patient and physician characteristics on the number of missing values in the study questionnaires.

Results: 72 out of 112 contacted dermatologists took part in the study (64.3%). There was an average time delay of four months between the first contact with dermatologists and the first contact with study subjects. Increased size of the catchment area (larger number of community codes the responding patients of each dermatologist belong to) corresponds with an increase of participation of physicians (OR 1.09 [KI 1.02-1.17]; p= 0.022). This effect also remains when only office-based dermatologists are examined (excluding hospitals): OR 1.08 [KI 1.01-1.17]; p= 0.032. Whereas age and cancer stage of eligible patients do not show any effects on the recruitment of the dermatologists.

Response proportions of patients depending on different study populations (subsets of the 2.113 identified study subjects) are
a) Contact proportion 1 (Percentage of patients for which there was a theoretical chance to contact them via the participating physician): 1.702/2.113=80.6%,
b) Contact proportion 2 (Percentage of patients finally contacted via the participating physician): 1.320/2.113=62.5%,
c) Cooperation proportion (Percentage of patients that participated of those who were contacted): 689/1.320= 52.2%,
d) Response proportion (Percentage of patients who participated of the total number of patients selected): 689/2.113=32.6%.

There were no relevant differences between participants and non-participants regarding site of tumor, time since diagnosis and place of residence (urban/rural). But there were slight differences regarding age distribution, sex and cancer stage: Non-participants more often belonged to older age groups and female sex. Furthermore they less often had an early tumor stage and more often unknown cancer stages respectively. No relevant differences considering age, sex, cancer stage, time since diagnosis, localization of melanoma or place of residence could be found between early responders (participants who sent back all study documents after the first mailing, n=401; 58.2%) and late responders (all other participants, n=288; 41.8%). However, a larger number of participant contact attempts corresponds to a larger proportion of missing values in questionnaires (OR 2.50 [1.28-4.85]). The same applies for elderly patients (OR 1.03 [1.01-1.04]) and female sex (OR 1.48 [1.03-2.12]).

The recruitment rate tends to be higher for medical practices than for hospitals. The median age of eligible patients of hospitals is higher than the median age of possible study subjects of medical practices (OR 1.11 [1.07-1.16]). But if participants are recruited by a clinic they do not differ in age from participants of office-based dermatologists. Patients invited to take part in the study by hospitals had to be contacted more often than patients invited by medical practices (OR 1.93 [1.27-2.93]. Residents of urban settings tend to be better reached if they are contacted by clinics (OR 1.75 [1.15-2.66]). Regarding data quality medical data provided by the cancer registry are better for participants recruited by hospitals, whereas there is no difference observed in the number of missing values in the study questionnaire between those and participants recruited by office-based physicians.

Discussion: Although the cooperation proportion of 52.2% in the MeLa-study is comparable to similar register based studies [2], [3], [4], [5] many potential study participants (n=411; 19.5% of the whole study population) were never contacted, because their treating/reporting physician did not take part in the study. The possibility of directly contacting eligible study participants by the cancer registry seems to lead to higher participation portions. Recruitment via office-based physicians has many disadvantages and is complex and expensive. However, if an indirect way of contact is mandatory, we recommend recruitment procedures including hospital- based rather than office-based physicians. Our experience is that by this way more patients could be reached with less effort. Moreover, data quality of medical information for patients recruited by hospitals was better in our study. In case of doubt, if study costs have to be reduced, it seems more effective to invest resources in recruitment of additional participants rather than repeated attempts of contact to responders who are undecided or very hesitant in participation.


References

1.
Slattery ML, Edwards SL, et al. Response rates among control subjects in case-control studies. Ann Epidemiol. 1995 May;5(3):245-9.
2.
Koch L, Jansen L, Herrmann A, et al. Quality of life in long-term breast cancer survivors – a 10-year longitudinal population-based study. Acta Oncologica. 2013 52:6, 1119-1128.
3.
Mehnert A, Koch U. Psychological comorbidity and health-related quality of life and its association with awareness, utilization, and need for psychosocial support in a cancer register-based sample of long-term breast cancer survivors. J Psychosom Res. 2008; 64:383-391.
4.
Waldmann A. Nolte S, Pritzkuleit R, et al. Different Aspects of Self-Reported Quality of Life in 450 German Melanoma Survivors. Cancers. 2011; 3: 2316-2332.
5.
Koch L, Bertram H, Eberle A, et al. Fear of recurrence in long-term breast cancer survivors-still an issue. Results on prevalence, determinants, and the association with quality of life and depression from the Cancer Survivorship-a multi-regional population-based study. Psychooncology. 2014; 23(5), 547-54. DOI: 510.1002/pon.3452 External link