Hum. Reprod. Advance Access originally published online on July 25, 2008
Human Reproduction 2008 23(11):2501-2505; doi:10.1093/humrep/den275
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
IVF patients show three types of online behaviour
1 Department of Obstetrics and Gynaecology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands 2 Department of Medical Psychology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands 3 Department of Medical Informatics, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands
4 Correspondence address. E-mail: w.tuil{at}obgyn.umcn.nl
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
BACKGROUND: The internet introduces new ways to deal with stress. However, it is unclear how its resources are used in everyday life. Using a web-based personal health record (PHR), we observed the patient’s online behaviour and linked this to distress, theories on dealing with stress and demographics.
METHODS: Between 2004 and 2007, all viewed web-pages were logged and categorized into 14 content types. Behavioural styles were elicited using factor analysis. These behavioural styles were subsequently correlated to data on demographics, coping mechanisms and distress from the female partner of the first 53 patient couples that used the PHR.
RESULTS: One thousand and fifty patient couples viewed 588 887 web pages during their first treatment cycle. Factor analysis elicited three online behavioural styles explaining 66.9% of all variance in usage of the website: an ‘individual information style’, a ‘generic information style’ and a ‘communication style’. The ‘individual information style’ correlated negatively to having paid employment (Spearman = –0.364, P = 0.007) and emotional coping mechanisms (Spearman = –0.305, P = 0.028). The ‘communication style’ correlated positively to having paid employment (Spearman = 0.318, P = 0.021) and anxiety (Spearman = 0.381, P = 0.005).
CONCLUSIONS: IVF patients show three types of online behaviour. Only limited correlations exist between these styles and demographics, coping mechanisms or distress. When planning a website or portal for IVF patients, content should be adopted accordingly.
Key words: IVF/coping/internet/consumer health informatics/factor analysis
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
An IVF treatment is a stressful period for many patients, which often results in increased depression and anxiety for the female partner after an unsuccessful treatment attempt (Verhaak et al., 2007
). Patients exhibit various coping mechanisms to deal with this stressful event (Peterson et al., 2006).
The internet introduces a new toolset for patients to deal with the problems in everyday life, including healthcare issues. It offers a vast resource for medical information, knowledge and decision aids (Eysenbach, 2000). Next to that, the internet facilitates peer-to-peer interaction via electronic support groups and virtual communities (Wellman, 1997). Various studies have shown that these contingencies are indeed used by a substantial portion of patients throughout the medical domain either to obtain information regarding healthcare issues or to communicate online about these issues (Hellawell et al., 2000; Ikemba et al., 2002; Metz et al., 2003). More specifically, the use of the internet for fertility-related searches in the IVF patient population is estimated at 42–54% (Haagen et al., 2003; Weissman et al., 2000), which may well have increased over the recent years. These figures show that the internet has become an important outlet for patients facing infertility.
In the non-medical domain, it is proposed that demographics do not predict online behaviour, but that lifestyle factors indicating a ‘wired’ lifestyle play a more dominant role (Bellman et al., 1999). Other factors like the time to first internet use, location of use and the type of applications used are also proposed as predictors of the frequency of internet usage (Emmanouilides and Hammond, 2000).
The effect of health-related internet use on emotional adjustment to health problems is not clear. Previous studies indicate that patients differ in the amount of information they need and in the need to participate in treatment decisions. It is suggested that information delivery should be adjusted to the individual needs of the patient and to the patient’s individual coping mechanisms (Miller and Mangan, 1983; Timmermans et al., 2007). Some patients tend to monitor health-related information and providing information to those patients is suggested to be beneficial for their emotional adjustment. Other patients, however, tend to avoid information. Those patients seem to adjust better by avoiding possible threatening health-related information. The internet could provide the patients the opportunity to confront themselves with as much information as they need; they are able to fully control the information supply by surfing or not surfing to specific sites. On the other hand, a patient centred internet-based programme could probably seduce patients to look for information, to confront themselves with information they would avoid in other media and that might elicit distress.
To date, no studies have observed the actual online behaviour of patients facing infertility or have identified factors that might explain or influence this behaviour. Unfortunately, current studies are limited to self-reports of overall internet usage. For example, one study in the IVF patient population reported ethnic background and annual household income to be significant predictors of overall internet use (Haagen et al., 2003).
At Radboud University Nijmegen Medical Centre, we developed in 2003 an Internet-based personal health record (PHR) that encompasses many of the online services provided across the Internet (Eysenbach, 2000). We have shown that this website is used by the majority of our patient population and that our patient couples value the functions it offers (Tuil et al., 2006). Moreover, we have shown that the website does not cause distress, but clear clinical benefits could not yet be demonstrated (Tuil et al., 2007). After an initial pilot phase, the website was adopted into routine care and was no longer treated as a novelty service. By logging all the couples’ behaviour, the website was a unique tool to unobtrusively monitor the patient couples' online behaviour.
The primary objective of this study was to explore the online behaviour of patient couples on the PHR website and to elicit behavioural styles that explain the variance in this behaviour. A secondary objective was to correlate these behavioural styles to a limited set of demographic characteristics, individual coping mechanisms, anxiety and depression collected for the female partner. The female partner was the focus of the study because previous studies in our clinic indicate that men did not show significant emotional responses to the IVF treatment irrespective of its outcome (Verhaak et al., 2005).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The interactive website developed at Radboud University Nijmegen Medical Centre offers a wide range of functions: patient couples were given access to all necessary medical information and were aided in interpreting this information. Next to that, the website aimed at establishing a virtual community and facilitated social networking both among patients and between patients and their physician. The website’s content was divided over a total of 14 content types, as shown in Table I.
|
Between January 2004 and December 2007, the website was used by 1150 patient couples. They all provided informed consent stating that their behaviour on the website was monitored and that these data could be used for scientific purposes. During that time, all pageviews were logged. For each viewed page, the name of the website’s user, the date and time the page was requested, and the name of the requested page were automatically logged and stored in the database. A patient couple was identified as a single user. As a result, data on usage cannot discriminate between the female and the male partner. On the basis of the name of the requested page, the pageviews were categorized into the 14 types of content (Table I). Pages used solely for navigation (i.e. having no actual content) were excluded from further analysis. Pageviews were included in the analysis if they occurred within a 70 day timeframe after a patient couples’ first visit to the website. This corresponded to the patient couples’ first IVF/ICSI treatment cycle, accounting for a 2 week time-lag in which the patient couples were not yet online, and a fact which can be mainly attributed to administrative procedures.
To gather data on demographics, individual coping mechanisms, distress in terms of anxiety and depression, the female partner of the first 53 patient couples who visited the website filled out a detailed written questionnaire. These 53 patient couples were the users of the website during the pilot phase of the project in 2004. The demographic variables in this survey were: age, nationality, employment, education level, household income and previous experience with IVF/ISCI treatments. Coping mechanisms were determined by the COPE (Carver et al., 1989). Four different coping mechanisms were assessed: active coping (including planning, suppression of competing activities, restraint coping); sharing emotions (including seeking emotional support, venting emotions); cognitive reinterpretation (including positive reinterpretation and growth, acceptance and behavioural disengagement) and denial (including mental disengagement). Furthermore, depression was measured through the Beck Depression Index for Primary Care (Beck et al., 1997) and finally, state anxiety was measured using the State-Trait Anxiety Inventory (Spielberger et al., 1970; Van Der Ploeg et al., 1980).
Factor analysis
To elicit the behavioural styles explaining the variance of usage on the website, the data on pageviews first had to be aggregated. This resulted in the number of pageviews per patient couple for the 14 types of content. Therefore, the aggregated data consisted of 14 variables per patient couple representing the website usage of that couple. The number of variables had to be reduced to elicit meaningful behavioural styles. We used factor analysis to accomplish this. Factor analysis is a statistical method used to explain variability among observed variables in terms of fewer unobserved variables. These fewer variables are called factors. We performed factor analysis using principal axis factoring as the extraction method. This method was also used by Zhou et al. in a similar context: they used this method to identify latent factors, which were used to generate aggregated user profiles for providing personalized recommendations. They have shown that this method is better at producing user profiles than factor analysis using principal component analysis or clustering methods (Zhou et al., 2004).
The Kaiser–Meyer–Olkin test and Bartlett test of sphericity were undertaken. These tests are used to establish the adequacy of the item correlation matrix upon which factor analysis is based. Extraction of factors was based upon Kaiser’s criterion for Eigenvalues of equal to or greater than unity.
To simplify the interpretation of the factors, varimax rotation with Kaiser normalization was used. Furthermore, regression scores were saved for the factors that resulted from the factor analysis. These factor scores are new variables that refer to a subset of the 14 content types representing online behaviour. Patient couples having a factor score above average were said to exhibit this behaviour. On the basis of this, the prevalence of the behaviours was calculated. The dominant behaviour was also determined. This was the behaviour having the highest score for that particular patient couple.
The regression scores for the latent factors were correlated to data from questionnaires filled out by the female partner of the first 53 patient couples that visited the website during the pilot phase in 2004.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The 1150 patient couples who visited the website viewed a total of 1 781 873 pages; 756 050 of those were during their first treatment cycle; and 167 163 of these pages (22%) were exclusively used for navigational purposes and excluded in the analysis. The remaining 588 887 pages (78%) had actual content and were used in the factor analysis. An overview of the types of pages and the number of pageviews is available in Table I.
The Kaiser–Meyer–Olkin coefficient for this data set was 0.916 and the Bartlett test of sphericity was statistically significant (
2 = 14 323.602, d.f. = 91, P < 0.0001) indicating that factor analysis can be carried out.
Factor analysis required 11 iterations and resulted in three factors explaining 66.9% of all variance. These three latent factors can be viewed as three types of online behaviour. The correlations between the 14 page types and the three behavioural styles resulting from the factor analysis are available in Table II. The first behavioural style accounted for 29.5% of all variation. It correlated to pages containing the individual information of the patient couples and is therefore called the ‘individual information style’. The page types this style correlated to were the day planner, the individual medical records, the downloadable files and the personalized prognosis. However, the ‘individual information style’ correlated also to the generic pages with frequently asked questions and treatment information, which were context-sensitive pages aiding in the interpretation of the medical record (i.e. these pages contained general information, but were only available through hyperlinks in the medical record). The second behavioural style accounted for 20.3% of the variation and correlated to pages containing generic information that was not context sensitive. Therefore, this style is called the ‘generic information style’. The style correlated to information concerning the IVF/ICSI treatment and our hospital, hyperlinks to external websites, recommended literature and help on how to use the website. The third behavioural style accounted for 17.2% of all variance and correlated to the communication functions of the website, i.e. the posting to and viewing of the discussion forum, and the chatroom. This behavioural style was called the ‘communication style’. The ‘individual information style’ was prevalent in 471 of all patients (41.0%), the ‘generic information style’ in 426 patients (37.0%) and the ‘communication style’ in 250 patient couples (21.7%). The prevalence of the different combinations of online behavioural styles is shown in Table III. The ‘individual information style’ was the dominant style in 33.2% of all patient couples; the ‘generic information style’ in 29.0% and the ‘communication style’ in 37.8%. Correlation to the demographic data revealed a negative correlation between the ‘individual information style’ and having paid employment (Spearman = –0.364; P = 0.007). For the ‘communication style’, there was a positive correlation with paid employment (Spearman = 0.318; P = 0.021). No other demographic variables correlated significantly. An overview is available in Table IV. Regarding the correlation with coping mechanisms, the ‘individual information style’ correlated negatively with emotional coping mechanisms (Spearman = –0.305; P = 0.028) indicating a negative relationship between sharing emotions and seeking individualist information on the website. In addition, the different online behavioural styles were not related significantly to depression. However, a trend could be observed for a relationship between depression and both the ‘individual information style’ (Spearman = –0.234; P = 0.092) and the ‘generic information style’ (Spearman = 0.237; P = 0.088). Moreover, the ‘communication style’ did correlate significantly to anxiety at the start of the IVF treatment (Spearman = 0.381; P = 0.005). An overview of these correlations is also available in Table IV.
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We could detect three online behavioural styles in IVF patient couples: an ‘individual information style’, a ‘generic information style’ and a ‘communication style’. There are only a limited number of correlations between these styles and demographics or coping mechanisms: patients having paid employment are significantly less attracted to individual information and significantly more to communication functions. Patients who exhibit the coping mechanism of sharing of emotions show less attraction to individual information. Although only a trend was observed, patients with a more depressive mood seem less likely to look for information concerning their own treatment and more likely to use the website to find generic information concerning the IVF treatment. A longitudinal study would explain this relationship in more detail. The positive relationship between anxiety at the start of the treatment and the ‘communication style’ suggests that more anxious patients are more likely to use the bulletin board (which provides asynchronous web communication) and the chatroom (for synchronous web communication) functions of the website.
It is possible that they communicate on the website as a way to relieve their anxiety. In a larger sample size, it would have been possible to investigate possible interaction effects between online behavioural style, coping mechanisms and anxiety and depression.
The lack of correlation between the behavioural styles and the demographic variables ‘household income’ and ‘ethnic background’ is notable. In previous research by Haagen et al. (2003), both variables were significant predictors of overall internet usage, whereas in the present study, they are not. This can be attributed to differences in both study populations. In contrast to our earlier observation, all participants in the present study had access to the Internet.
Thirty-seven per cent of the patient population does not exhibit any of the three styles. This subgroup can be characterized as the non-users: they have signed up for the website, but have only used it on limited occasions. There are three explanations for this: first, this subpopulation has no need for online support during the IVF/ICSI treatment. Second, this subpopulation does not have the opportunity to use the website, e.g. due to lack of time. Third, the content of the patient portal does not yet suit the needs of this subpopulation. If the latter is the case, this would be an incentive to add more content to the patient portal. However, during 3 years of operation, we did not get any indications of missing content.
The data resulting in the three types of online behaviour were gathered for patient couples. We assumed that when using the website, patient couples would share their logon credentials out of convenience. In order to reduce the complexity of the logon procedure of the PHR, we only supplied one set of login credentials to the female and the male partner. Undoubtedly, there are gender differences in usage between partners of the patient couple. These gender differences have been observed for the participation in online communities (Ginossar, 2008) and also on coping mechanisms for patients facing infertility (Peterson et al., 2006). However, due to the fact that female and male participants shared their logon credentials, this distinction could not be made in our analysis. Results from a previous study, however, revealed that in 72% of all cases, the female partner used the website the most; in 20%, the partners were equally active and in only 8%, the male partner was more active (Tuil et al., 2006). Therefore, the results from the present study probably predominantly apply to the female partner of the IVF patient couple.
Next to that, the three behavioural styles are based on aggregated data of the pages viewed over a 3 month period. The usage of the website fluctuates over this period (Tuil et al., 2008) and therefore the expression of the three styles probably is not constant. This study, however, cannot give insight into the relationship between time course and expression of the styles.
Last, it is debatable whether the correlations between the behavioural styles on the one hand and demographics, coping mechanisms, anxiety and depression on the other have enough statistical power. This analysis was done based on data from a subpopulation of 53 female patients who used the website during its pilot phase in 2004. This is a relatively small population, which results in relatively low statistical power. However, for the statistically significant correlations found in this study, the statistical power (
= 0.05, two-tail) lies between 0.60 (for the relationship between sharing emotions and the ‘individual information style’) and 0.81 (for the relationship between anxiety at the start of treatment and the ‘communication style’). Although a power of 0.60 may be on the low side, a power of 0.81 is quite adequate.
The three styles could of course be used as a profile to base website personalization on. More importantly, however, one has to recognize that these three styles exist. Some patients will, therefore, be more attracted to individual information, general information and communication facilities. Healthcare providers who want to provide internet services for their patient population should therefore incorporate all three content types.
Furthermore, our study shows that traditional demographics are not good at predicting the actual usage of internet services. Moreover, traditional coping mechanisms do not predict the patients’ online behaviour either. Therefore, traditional demographics are not likely to be valuable covariates in Internet research. Subsequently, a new evaluation framework needs to be developed that focuses on lifestyle factors and patient characteristics rather than on demographics.
Finally, anxious patients are more likely to use communication modalities on the Internet, such as bulletin boards or chatrooms. These applications may function as an outlet for negative emotions and offer relief in stressful periods. Moreover, these types of applications offer new opportunities for interventions aimed at the reduction of anxiety.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Beck AT, Guth D, Steer RA, Ball R. Screening for major depression disorders in medical inpatients with the Beck Depression Inventory for Primary Care. Behav Res Ther (1997) 35:785–791.[CrossRef][Web of Science][Medline]
Bellman S, Gerald LL, Eric JJ. Predictors of online buying behavior. Commun ACM (1999) 42:32–38.
Carver CS, Scheier MF, Weintraub JK. Assessing coping strategies: a theoretically based approach. J Pers Soc Psychol (1989) 56:267–283.[CrossRef][Web of Science][Medline]
Emmanouilides C, Hammond K. Internet usage: Predictors of active users and frequency of use. J Interact Mark (2000) 14:17–32.[CrossRef]
Eysenbach G. Consumer health informatics. BMJ (2000) 320:1713–1716.
Ginossar T. Online participation: a content analysis of differences in utilization of two online cancer communities by men and women, patients and family members. Health Commun (2008) 23:1–12.[CrossRef][Web of Science][Medline]
Haagen EC, Tuil W, Hendriks J, de Bruijn RP, Braat DD, Kremer JA. Current Internet use and preferences of IVF and ICSI patients. Hum Reprod (2003) 18:2073–2078.
Hellawell GO, Turner KJ, Le Monnier KJ, Brewster SF. Urology and the Internet: an evaluation of internet use by urology patients and of information available on urological topics. BJU Int (2000) 86:191–194.[CrossRef][Web of Science][Medline]
Ikemba CM, Kozinetz CA, Feltes TF, Fraser CD Jr, McKenzie ED, Shah N, Mott AR. Internet use in families with children requiring cardiac surgery for congenital heart disease. Pediatrics (2002) 109:419–422.
Metz JM, Devine P, DeNittis A, Jones H, Hampshire M, Goldwein J, Whittington R. A multi-institutional study of Internet utilization by radiation oncology patients. Int J Radiat Oncol Biol Phys (2003) 56:1201–1205.[CrossRef][Web of Science][Medline]
Miller SM, Mangan CE. Interacting effects of information and coping style in adapting to gynecologic stress: should the doctor tell all? J Pers Soc Psychol (1983) 45:223–236.[CrossRef][Web of Science][Medline]
Peterson BD, Newton CR, Rosen KH, Skaggs GE. Gender differences in how men and women who are referred for IVF cope with infertility stress. Hum Reprod (2006) 21:2443–2449.
Spielberger C, Gorsuch R, Lushene R. STAI Manual for the State-Trait Anxiety Inventory (1970) Palo Alto, CA: Consulting Psychologists.
Timmermans LM, van Zuuren FJ, van der Maazen RW, Leer JW, Kraaimaat FW. Monitoring and blunting in palliative and curative radiotherapy consultations. Psychooncology (2007) 16:1111–1120.[CrossRef][Medline]
Tuil WS, Ten Hoopen AJ, Braat DDM, de Vries Robbé PF, Kremer JAM. Patient-centred care: using online personal medical records in IVF practice. Hum Reprod (2006) 21:2955–2959.
Tuil WS, Verhaak CM, Braat DD, de Vries Robbe PF, Kremer JA. Empowering patients undergoing in vitro fertilization by providing Internet access to medical data. Fertil Steril (2007) 88:361–368.[CrossRef][Web of Science][Medline]
Tuil WS, van Selm M, Verhaak CM, de Vries Robbe PF, Kremer JA. Dynamics of Internet usage during the stages of in vitro fertilization. Fertil Steril (2008) April 28 (Epub ahead of print).
Van Der Ploeg H, Defares P, Spielberger C. Handleiding bij de Zelf-Beoordelings Vragenlijst ZBV: Een nederlandstalige bewerking Van de Spielberger State-Trait Anxiety Inventory, STAI-DY [Manual of the Self-Evaluation Questionnaire: a Dutch version of the State-Trait Anxiety Inventory]. (1980) Lisse, The Netherlands: Swets & Zeitlinger B.V.
Verhaak CM, Smeenk JM, van MA, Kremer JA, Kraaimaat FW. A longitudinal, prospective study on emotional adjustment before, during and after consecutive fertility treatment cycles. Hum Reprod (2005) 22:2253–2260.
Verhaak CM, Smeenk JM, Evers AW, Kremer JA, Kraaimaat FW, Braat DD. Women’s emotional adjustment to IVF: a systematic review of 25 years of research. Hum Reprod Update (2007) 13:27–36.
Weissman A, Gotlieb L, Ward S, Greenblatt E, Casper RF. Use of the internet by infertile couples. Fertil Steril (2000) 73:1179–1182.[CrossRef][Web of Science][Medline]
Wellman B. An electronic group is virtually a social network. In: Cultures of the Internet—Kiesler S, ed. (1997) Mahwah, NJ: Lawrence Erlbaum. 170–205.
Zhou Y, Jin X, Mobasher B. A recommendation model based on latent principle factors in web navigation data. Proceedings of the 3rd International Workshop on WebDynamics at WWW 2004 Conference, 2004: New York.
Submitted on March 10, 2008; resubmitted on June 13, 2008; accepted on June 18, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Chiu, P. Ande, R. A. Coward, and A. Woywodt The times they are a changin'--the Internet and how it affects daily practice in nephrology NDT Plus, August 1, 2009; 2(4): 273 - 277. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||