i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786

journa l homepage: www. int l .e lsev ierhea l th .com/ journa ls / i jmi

Access to and use of the Internet by South Africangeneral practitioners

Ken Masters ∗

ITHealthEd, Köstendorf, Austria

a r t i c l e i n f o

Article history:

Received 21 August 2007

Received in revised form

31 March 2008

Accepted 27 May 2008

Keywords:

Internet

Family practice

Health care surveys

South Africa

Developing countries

a b s t r a c t

Introduction: In the developed world, doctors use the Internet to support the delivery of healthcare. Their usage patterns are explained by Rogers’ Diffusion of Innovations. There is no

knowledge of Internet usage by doctors in the developing world.

Methods: In July 2006, 2600 survey forms were posted in the first national survey of South

African (SA) General Practitioners’ (GPs) use of the Internet. A sample of non-responders

was followed up. The aim was to determine SA GPs’ Internet usage patterns, to compare

them to world usage, and to examine them in the light of Diffusion of Innovations.

Results: A 10% usable response rate was a major weakness, but similar surveys are unlikely

to be more representative. 89% of SA GPs have Internet access, home usage is high, and

overall usage patterns are equivalent to the usage patterns of international studies.

Discussion: In spite of the overall technological under-development in South Africa, as pre-

dicted by Diffusion of Innovations, SA GPs have adopted the technology, and use it for health

care in much the same way as their international counterparts. Further studies on some of

the details will be valuable.

Conclusion: The indications are that SA GPs will reap the benefits of the Internet as it con-

tinues to evolve, and translate these into improved health care delivery in South Africa.

In that review, the results were discussed in the light of Rogers’

1. Introduction

Numerous publications discuss the need for doctors tomaintain expertise through formal and informal ContinuingMedical Education (CME) (e.g. [1,2]), to remain in contact withcolleagues (e.g. [3,4]) and also to maintain good communica-tion with patients (e.g. [5,6]).

Internationally, doctors’ use of the Internet has been stud-ied widely, by both independent researchers and medicalassociations, in the USA, Canada, and Europe [7]. A systematicliterature review of these and other studies conducted by the

author [7] revealed that, on average, doctors’ use of the Inter-net is higher than their national averages, and is increasing.Doctors use the Internet for general medical research and per-

∗ Tel.: +43 664 4982 076.E-mail address: kmasters@ithealthed.com.

1386-5056/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights resdoi:10.1016/j.ijmedinf.2008.05.008

© 2008 Elsevier Ireland Ltd. All rights reserved.

sonal use, accessing a wide range of sites including journals,CME sites, financial sites, and, to a lesser extent, to maintaincontact with colleagues and patients. Males tend to use theInternet more than females, and young more than old, butthese figures are not consistent across all studies, and appearto be equalising.

1.1. Theoretical background

Diffusion of Innovations (DoI) [8,9]. DoI was able to explainhow the Internet’s ability to meet doctors’ professional needsfor high quality health care delivery drove their adoption of

erved.

a l i n

It

nfbts

smstht

i“aasiaaste

t

1

TticZtiwa

ktrGt

1

Wicu(Ee[ta

i n t e r n a t i o n a l j o u r n a l o f m e d i c

nternet technology. It is worthwhile to describe some of theheoretical details here.

Rogers’ work on DoI was first published in the 1960s, and isow in its fifth edition [9]. In 1997, Rogers and Scott gave the

ollowing description of Diffusion, saying that it is the “processy which (1) an innovation (2) is communicated through cer-ain channels (3) over time (4) among the members of a socialystem” [8,9]. These four elements form the basis of DoI.

In his writings, Rogers often refers the “members of aocial system” as “units”, which might be “individuals, infor-al groups, organisations, and/or subsystems” [8]. The social

ystem, not limited to a particular geographical area, “consti-utes a boundary within which an innovation diffuses” andas norms, which are “the established behavior patterns forhe members of a social system” [8].

The diffusion of technology across a social system isncreased when the social system consists of similar, orhomophilous” [9], individuals. Rogers defines “homophilous”s “the degree to which two or more individuals who inter-ct are similar in certain attributes, such as beliefs, education,ocio-economic status, and the like” [9]. Their common groundncludes education, membership of professional bodies, and

“mutual subcultural language” [9]. DoI also argues that thedoption of technology by members of this social group istrongly influenced by the technologies’ usefulness and valueo them, and the demands of the social environment andxpectations of their clients.

Finally, DoI points to a strong correlation between educa-ional levels, social status, and adoption of technologies [9].

.2. The theory and this paper

he issues raised by DoI are crucial here, because this paperries to rectify a large weakness that was identified in that orig-nal literature review. There were no studies from developingountries, and only one from the Southern Hemisphere (Newealand) [10]. The principles of DoI predict that it is possiblehat the Internet could play a similar useful role in health caren developing countries. For this to be established, however,e need to know more about the patterns of Internet usagemongst doctors in those countries.

This paper aims at closing some of that gap in ournowledge by examining South African medical general prac-itioners’ (GPs’) use of the Internet. This will be done byeporting on the results of a national survey of South AfricanPs, and examining these results in the light of the interna-

ional literature and Diffusion of Innovations.

.3. South Africa

hile South Africa (SA) is not a typical developing country,t is certainly not a developed country, and SA’s technologi-al infrastructure is variable. Only 9% of the SA populationses the Internet [11], and its Economist Intelligence Unit’s

EIU) 2006 e-readiness ranking [12] is far below that of Westernurope and the USA. That said, however, its EIU is the high-

st in Africa, is higher than many Eastern European countries12], and its mobile-phone penetration of 71.6% is equivalento that of the USA (71.4%), and more than double the worldverage (34%) [11].

f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786 779

After training for seven years, SA GPs practice generalmedicine in this diverse environment, with its wide-rangingdifferences in patient access to health services amongsturban, rural and peri-urban “informal settlements” (In SA,there is no accepted definition of these) [13]. In addition, thereis a disparity in service provision according to setting, whetherthe GP is in private general practice (PGP), a public hospital(HGP) or a public clinic (CGP).

In spite of this variability, South African GPs are part ofthe international medical fraternity, undergoing internation-ally recognised training, and maintaining their qualificationsthrough formal and informal CME. SA GPs’ conform to inter-national standards, access research, communicate with peers,share information and follow similar medical practices. They,therefore, form part of the international medical “socialgroup,” and it is reasonable, at this stage, to propose that theiradoption of technology is surely to be influenced by the samefactors that influence their peers, and will be at a comparablerate of adoption.

In addition, with approximately only 10% of South Africans’having a university degree [14], it is to be expected that GPswith at least 6 years’ university training do not represent theoverall educational level of the SA population. If the predi-cations of DoI are broadly applicable, then one would expectthe GPs’ adoption of new technologies to be disproportion-ally higher than the average for the population, and relativelycomparable to that in developed countries.

1.4. Aim of this paper

While we are aware that SA’s GPs are using the Internet, wehave no accurate information on their usage. This paper aimsto supply that information, and uses the first national surveyto do so.

In the process, this paper will attempt to answer 3 ques-tions:

• What are the overall patterns of Internet usage by SouthAfrican GPs?

• How do the Internet usage patterns compare to interna-tional usage?

• What is the impact of the demographic characteristics onSA GPs’ Internet usage patterns?

2. Methods

2.1. Sampling and methodology

A questionnaire, based on the systematic literature review [7],was designed. After ethics approval from the University ofCape Town’s (UCT) Faculty of Health Sciences’ Ethics’ Com-mittee, the original questionnaire was piloted on five local GPs,modified appropriately, and revised to finally contain 39 items.

The target population was all GPs practising in SouthAfrica. An ideal sample size, for examination of small sub-

groups, was statistically determined to be 1300. In July 2006,optimistically aiming for a 50% return, questionnaires weremailed to 2600 names drawn from the South African Medi-cal Association’s (SAMA) database of GPs. SAMA is the largest

i c a l i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786

access, access from work and access from home are all sta-tistically greater (p < 0.05) than the figures in the internationalreview [7]. Part of the reason lies in the methods of access fromtheir work place, shown in Table 2.

Table 1 – Place and percentage of access

Location % of GPs accessing itfrom each Location

Relative % of timespent on the Internet

(n = 225)

780 i n t e r n a t i o n a l j o u r n a l o f m e d

medical union in South Africa, with more than 17,000 GPs inits database. This was a random sample, stratified into 70%metropolitan, 30% non-metropolitan, matching the propor-tions in the SAMA database. The questionnaire pack includeda letter, brochure, consent form, a pre-paid self-addressedenvelope, and eligibility of entrance into a draw to win oneof three prizes worth approximately $70 each. Respondentswere also given a URL through which they could completethe survey online. Financial constraint prevented follow-uppostings.

Although the letter emphasised that non-Internet usersshould also complete the survey, there was a risk that theywould not, and that the percentage of users would, therefore,be skewed. The hypothesis that Internet users would be dis-proportionately represented amongst the respondents wouldhave to be tested. A random selection of 50 non-respondentswould be selected, and interviewed telephonically on theirInternet usage, email with patients, and their reasons for notresponding. If this group were statistically consistent with thesample, then this hypothesis could be rejected.

3. Results

3.1. Non-responders

Of the randomly selected 50 non-responders, only 14 could becontacted. Only two did not use the Internet, and one usedemail with patients. Each non-responder did not rememberreceiving the questionnaire, and four said that if they had,they probably would not have completed it because of lackof time. These results are discussed later in the light of theoverall statistics to follow.

3.2. Overall response rate and validation

A total of 394 (15.2%) responses were received. 107 werereturned as undeliverable, and 28 were returned by doctorsno longer working as GPs. Of the 259 (9.98%) valid responses,19 were received online.

Although 259 is a reasonable absolute number, it is alow response rate, and raises a problem of the samples’representivity. Because of this, the sample’s demographiccharacteristics were compared to those of the SAMA database.Comparisons were made on gender, age, and location, as theseare frequently identified as possible predictors of Internetusage in the literature [7].

3.3. Demographics

Of the sample, 68.9% (63.1–74.4) of the GPs were male. TheSAMA database has 65.4% male. A �2 value of 1.48 (p = 0.2244)was returned indicating no statistical difference between thesample and the SAMA database.

The mean age in the sample was 43.7 (42.3–45.2), and themedian was 43, while in the SAMA database, the mean age

was 40.5, indicating that the mean age of the sample was sig-nificantly higher than the mean age of the SAMA database.For further comparison with the SAMA database, the ageswere classified into 12 groups and displayed on a bar chart

Fig. 1 – The percentage of respondents in age groups vs.the percentage of GPs in the SAMA database, per age group.

(Fig. 1), which shows graphically that the sample has an under-representation in the younger age groups. The reasons andimplications of this are explored in the Discussion below. Forpurposes of reporting, the sample was divided into three agegroups: <35 (27.0%); 35–50 (45.2%); >50 (27.8%).

The sample’s geographical location (based on postal codes)was categorised according to SA’s nine provinces, and a Spear-man Rank Correlation against the SAMA database was 0.937,indicating a very strong correlation of geographic spreadacross the country.

GPs were classified according to their location (urban,rural, peri-urban) category if they spent at least 50% oftheir professional time there. 78.4% (74.4–83.4) were urban,15.8% (11.4–20.3) rural, and 4.6% (2.1–7.2) peri-urban. Thiscompares closely to the 70/30 split of metropolitan and non-metropolitan categories of the SAMA database.

61.0% (55.1–66.9) of the GPs were in private general practice,23.2% (18.0–28.3) in public hospitals, 7.0% (3.9–10.0) in publicclinics, and the rest spread amongst company medical units,academic institutions, and other facilities. Of the private GPs,70.8% were in practices of two or fewer doctors.

3.4. General access and usage patterns

3.4.1. Internet accessIn total, 89.19% (85.4–93.0) of the sample access the Internet.Table 1 shows the location of Internet access and relative timethese doctors spend on the Internet.

From Table 1, we see that access from home is highcompared to other locations. Indeed, the figures for Internet

Home 82.2 53.1Clinical practice 56.9 32.4Hospital 16.9 6.9Other 11.6 6.9Unknown 0.7

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786 781

Table 2 – Method of Internet access at work

Activity % Access (n = 259)

Dialup (through a 56 kb modem) 50.2Broadband 18.6A network connection 35.8

pcta

3Omawt

mtinh

3Bolii(wwaa

iifair

i

Table 4 – Percentages of SA GPs’ involvement inactivities on the Internet, grouped into categories

Activity SA % LR % p

EmailPersonal email 86.5 82 0.0708Professional email to colleagues 40.5 24 <0.0001Professional email to patients 21.6 22 0.1107

Professional—clinicalSearching for drug information 46.0 37 0.0044Searching for patient-specific

information (diagnostic assistance)43.2 44 0.8344

Obtaining/transferring laboratoryresults

18.9 20 0.5575

Looking for sites to recommend topatients

17.4 24 0.0115

Participating in clinical trials 8.1 N/AObtaining/transferring medical

records6.2 N/A

Filling prescriptions/orders 5.8 N/A

Professional—non-clinicalVisiting professional bodies’ websites 64.5 27 <0.0001Communication with professional

bodies36.3 N/A

Filing medical aid/insurance claims 34.4 12 <0.0001

Personal (excludes email)Online banking 64.5 N/ATravel information/arrangements 58.3 39 <0.0001Entertainment 53.3 N/APersonal purchasing/shopping 42.1 N/ACollecting financial news 27.4 34 0.0332Internet telephone (VoIP) 10.4 N/A

Education, professional developmentReading online journals 54.8 52 0.3639Attend online CME

courses/conferences24.3 30 0.0408

Collecting conference information 19.3 N/ATeaching 9.7 N/AAttend online conferences 4.3 N/ATelemedicine 3.5 N/A

Other 6.2

For ease of comparison with the literature review [7], percentagesare given as a percentage of the total sample, and the percent-ages from the literature review are repeated here. The figures arecompared for statistically significant differences (p < 0.05), and thep values given; significant p values are in bold. Activities not listed

Laptop (WiFi/GPRS, etc.) 15.3Other Mobile Device (e.g., PDA, cell phone) 9.3Do not know 7.0

While access from work is high, Table 2 shows that therime method of access from work is through a 56 kb dialuponnection. In SA, where local calls are charged per minute,his is both expensive and inefficient. This poor access prob-bly encourages doctors to have home access.

.4.2. Amount of Internet usagef the Internet users, 76.1% had been using it for 5 years orore, and 24.4% had been using it for 10 years or more. 51.3%

ccessed it daily, and 33.5% 2–3 times per week, 7.8% once pereek. Table 3 shows the number of hours per week spent on

he Internet.44.1% said that their usage had increased over the past 6

onths, 52.0% said it had remained the same, and 3.9% saidhat it had decreased. 74% said that it improved or significantlymproved their practice of medicine, while 25% said it had hado impact. Only one GP in the sample felt it had worsenedis/her practice of medicine.

.4.3. Activities on the Internetased upon the activities raised in the literature review [7], andther activities that had been raised in some of the papers, a

ist of 26 activities was created. Respondents were asked tondicate whether they use the Internet for any of these activ-ties (“For which of these activities do you use the Internetindicate all that apply)?)”. This question emphasised that itas the activity of the GP personally; so that if an activityas performed in practice by an administrative staff member,

nd not by the GP, then it was not included. 231 respondentsnswered the question, and the results are given in Table 4.

Expectedly, email usage is high, and matches or exceedsnternational norms. Professional use generally matchesnternational norms. The low percentage of doctors lookingor sites to recommend to patients is explained by the lowccess to the Internet that the general SA population has, and

mpacts directly on the informed patient in the patient–doctorelationship.

At a first glance, the high score of personal usage, includ-ng entertainment, might be worrying, as it might conjure an

Table 3 – Total number of hours per week spent on theInternet (n = 230)

Number of hours Percentage of users

1–4 h 48.75–10 h 27.011–15 h 12.616–20 h 6.121 h or more 5.7

in the review are labelled “N/A”.

image of a GP in the office all day performing personal work.Given, however, that GPs access the Internet more from homethan from any other location, a high personal use is to beexpected.

Overall, where categories are common to both the reviewand this study, the usage by SA GPs is generally either theequivalent to or greater than the international level. Excep-tions are Financial News (although online banking is high at64.5%), attending online CME courses, and looking for sitesto recommend to patents. Given that SA GPs are expected to

attend CME courses, the low CME attendance online will placea greater burden on them, as they need to travel to attend CMEactivities.

782 i n t e r n a t i o n a l j o u r n a l o f m e d i c a l

Table 5 – Percentages of SA GPs’ visiting specificwebsites within the past 3 months

Website Percentage

Medical AssociationSAMA 33.6Health Professionals’ Council of

South Africa (HPCSA)32.4

Independent Practice Assoc (IPA) 10.8

General search engineGoogle (for searching) 75.3Google Scholar 7.7Yahoo (for searching) 34.0Another search engine 13.5

Medical—clinicalMedline 29.3PubMed 26.6Medscape 13.5Physicians Online 4.3WebMD 4.3MD Consult 3.9Other medical sites 24.7

Medical—non-clinicalMedical aid site 32.1Discovery Healtha 29.3A Pharmaceutical Co. 5.8

Newspapers and news stationsAny online newspaper 29.0CNN 8.5Other foreign news station 8.9MNetb 7.7SABCb 4.6ETVb 1.9

Other personal sitesYour bank 61.8An airline 44.8A travel site 42.5Amazon 23.9Cell phone/provider 17.0Telkom 12.7

Other 5.0No Web pages visited 2.3

Percentages are given as a percentage of the total sample.

a An SA Medical Aid Site.b A local television station.

3.4.4. Websites visitedBased on the websites examined in the literature, and addingother local sites, a list of 29 Websites (plus a “None” and“Other”), was created. 228 respondents answered the ques-tion identifying the Websites that they had visited in the past3 months. Table 5 shows the responses to this question.

As the source of the addresses is the SAMA Database, onemight expect the accesses to that site to be higher than it is.Given, however, that all South African doctors, irrespectiveof their medical affiliations, are members of the Health Pro-fessionals’ Council of South Africa (HPCSA), the visits to the

SAMA are relatively high.

Google is the most commonly viewed search engine, and, infact, overall the most commonly viewed site. This, especiallyin the light of the fact that accesses to Google Scholar is so

i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786

low, is probably a result of the Google search engine’s being astandard component on many peoples’ browser window, andgeneral familiarity. This should be examined further.

Although there is low access to sites like Medscape, Physi-cians Online, WebMD, and MD Consult, the sites of Medlineand PubMed have significantly more access to them. As thesesites concentrate on traditional and peer-reviewed journals, itindicates that most doctors are hesitant to move away fromthese.

The Medical Aid sites would be of particular importance todoctors in private practice, so the relatively high number ofaccesses is consistent with the high percentage of private GPsin the sample (61%).

The personal sites, especially the bank and airline, are con-sistent with the activities in Table 4, also reflecting the impactof the large amount of access from home (Table 1).

Overall, it is obvious that the Internet is part of doctors’personal and professional life; again emphasised by the largeamount of access from home.

3.5. Relationships between variables

The literature review had commented on the statisticalvariations of activities according to different demographiccharacteristics. To determine whether there were significantdifferences along the demographic characteristics in this sam-ple, the data were analysed with Bonferroni and �2 tests. In theanalysis below, only variables with figures that show statisti-cal differences (p < 0.05) are listed. While other variations mayoccur, they are too small to be statistically significant.

3.5.1. AgeThe relationship between the three age groupings (<35; 35–50;>50), and amount of Internet usage was complex. Overall, agreater percentage of younger doctors used the Internet, andused it from hospital, and used it for more activities. They alsovisited Medline, Pubmed, WebMD, Google, Yahoo, and theirBank more.

There were many times, however, where the middle agegroup had the highest usage. These were for Medscape, per-sonal email, professional email to colleagues, online banking,visiting professional bodies’ websites, entertainment, search-ing for patient-specific information, and looking for sites torecommend to patients.

There was a surprising difference in the daily access tothe Internet; a greater percentage of the older GPs used it ona daily basis, and visited Independent Practice Associations(IPAs) more. Fewer older GPs said that their use of the Inter-net had improved or significantly improved their practice ofmedicine.

It appears, then, that pattern does emerge where theyounger GPs perceive greater value from the Internet, are morecomfortable with Medical websites, banking, and also as partof a common ground in the relationship with patients. Thereare, however, similarities of usage, especially amongst the twogroups below the age of 50.

3.5.2. GenderIn spite of there being no difference between the percentageof male GPs and percentage of female GPs who were Internet

a l i n

untta

foa

yiop

fsPn

aIpaf

3Btdathwwf

deyst

GGsmt

tufa

3hMaaa

i n t e r n a t i o n a l j o u r n a l o f m e d i c

sers, a greater percentage of males had access to the Inter-et from work, and a greater percentage of males had accesshrough 56 kb dialup. A greater percentage of males accessedhe Internet from their clinical practice, and spent more timeccessing it from their clinical practice.

It appears that females make up for their lack of accessrom work by accessing the Internet from home: of their timen the Internet, females spent a greater percentage of timeccessing the Internet from home, than males did.

More males than females had been using the Internet for 10ears or more and 5 years or more, and more males also usedt on a daily basis. More males than females said that their usef the Internet had improved or significantly improved theirractice of medicine.

Males engaged in some Internet activities more thanemales: medical aid/insurance, financial news, teaching, per-onal purchasing/shopping and three web sites: Independentractice Association, cell phone provider, and other foreignews station.

Overall, then, there is no difference between the percent-ge of male and female users, but males have been using thenternet for longer, generally still use the Internet more, anderceive greater value from the Internet. Some of the figuresre equalising, however, and females use the Internet morerom home than males do.

.5.3. Location (urban, rural, peri-urban)ecause the peri-urban sample was so small [12], it is difficulto determine the statistical significance of the figures showingifferences between urban, rural and peri-urban. There werefew non-statistical variations, such as the fact that 100% of

he peri-urban doctors who accessed the Internet, did so fromome, whereas the mean for home access amongst all usersas 82% (p = .297) which leads one to suspect that those whoork in the peri-urban, live elsewhere, and access the Internet

rom home.For the most part, however, those differences in usage that

id occur had the pattern of urban > rural ≥ peri-urban. Thisxists for percentage of GPs using the Internet for more than 5ears, using it on a daily basis, the number of hours per weekpent on the Internet, use of email to patients, and participa-ion in clinical trials.

There were also other activities in which the peri-urbanPs did not participate at all, but the figures for the otherPs in these activities were too low to draw significant conclu-ions. These were transferring laboratory results, transferringedial records, prescriptions, looking for sites to recommend

o patients, and telemedicine.For these doctors, then, the variations of access still point

o a high degree of accessibility, with the suggestion that peri-rban doctors making up for frequency of access with access

rom home. Patterns of usage reflect primarily the medicalctivities in those locations.

.5.4. Setting (private general practice (PGP) vs. publicospital (HGP) and public clinics (CGP))

ore PGPs and HGPs had access from work than CGPs,

lthough actual usage differed. Specifically, more PGPsccessed the Internet from their practice, while more HGPsccessed it from other places, such as Internet cafés.

f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786 783

Often, usage followed a pattern similar to that of the Loca-tion, with PGP > HGP ≥ CGP. This included the percentage ofGPs using the Internet on a daily basis, email to patients, filingmedical aid and insurance claims, prescriptions, participatingin clinical trials, online banking and Discovery Health.

There were, however, a few variations. HGPs had great-est use of the Internet for email to colleagues, and teaching,and they were comparable to PGPs in the use of the Internetfor entertainment, and accessing Google. In two instances,the usage patterns of CGPs and HGPs (visits to medical aidsites and IPAs) were reversed. Reverse differences existed forPubMed and Medline, with CGP accessing these sites morethan the HGP and PGPs.

Related to the pattern of access from work, more PGPs andHGPs indicated that the Internet had improved or significantlyimproved their practice of medicine.

Overall, then, many of the usage patterns resemble the splitalong lines of urban, rural and peri-urban, and their patternsof usage reflect the medical activities in those locations, withthe CGPs making use of home access, and accessing the Inter-net less frequently. Because HGPs’ access from other places ishigher than the other two groupings, this suggests that theirneed for access is as strong, but that they have to compensatefor poor access facilities in the clinics.

3.5.5. Practice size of private general practicePrivate General Practices were classified into small (1–2 practi-tioners) and large (more than 2 practitioners). Larger practiceshad greater access from work and more network connectiv-ity, while smaller practices had greater access through 56 kbdialup.

The only differences in activities were travel information,with doctors in larger practices accessing this more, and filingof medical aid and insurance claims, with these doctors doingthis less.

Consistent with these activities, doctors in smaller prac-tices visited Discovery Health, medical aid sites and SAMAmore, and also visited travel sites and airlines less.

Overall, then, an important difference between small andlarge practices is method of access. In addition, it appears thatdoctors in larger practices perform fewer administrative taskson the Internet (perhaps having this performed by adminis-trative staff), and they appear to travel more.

4. Discussion

4.1. Comparison with international figures

A question asked by this paper was “How do the Internet usagepatterns in SA compare to international usage?” We shouldremember that the literature review spanned 10 years, andthis survey was conducted in 2006. That said, however, theresults indicate SA GPs’ degree of access to and use of theInternet is at least equivalent to international levels, implyingthat DoI has correctly predicted the SA GPs’ usage of the Inter-

net. This would argue that the motivating factors for GPs’ useof the Internet are similar to their international peers – thatthey have recognised the advantages of the Internet as a toolto assist with health care, that they are part of an international

i c a l

784 i n t e r n a t i o n a l j o u r n a l o f m e d

medical fraternity, conforming to international practise, andare influenced by the same factors that affect their interna-tional peers.

In DoI terms, they conform to the standards of an interna-tional social system [9], of which they are a part. They need tocommunicate with other doctors, both nationally and inter-nationally, with whom they are homophilous, and use it inways that are similar to their international counterparts. SAGPs, therefore, require access to the Internet, and this require-ment appears strong enough to overcome the financial andtechnological barriers that might exist in SA.

This need to access the Internet is emphasised by theiraccess patterns. Although access to the Internet from workis high, it is clear that there are barriers preventing adequateusage from work. These might plausibly be the poor quality ofaccess (56 kb dialup), and time available, as these have beenfound in other studies [7,15]. To counter this, GPs access theInternet from home to an extent that is higher than foundin other studies [7,16], at great expense [12], or from mobiledevices. This is particularly reflected in the high home accessof females and the tendencies in the CGPs and peri-urban GPs.

4.2. Demographic factors

The impact of the demographics factors, with unclear bound-aries, resembles the findings in the international literature,and is also predicted by DoI [9]. The traditional patterns exist,but, in many instances, show signs of equalising. For exam-ple, just as there is a blurring of groups below the age of 50,one can expect that, over the next 10 years, this blurring willoccur throughout the GP population groups. While groups willalways have preferences for particular sites, it is unlikely to bevery pronounced.

4.3. Limitations of the study

An obvious weakness of the study is the low response rate.This low response rate to postal surveys amongst doctors is aproblem internationally [17,18], usually because of lack of timeor overload of work. Other national postal surveys of doctorshave received response rate of less than 30% (e.g. [19,20]), whileothers conducted via telephone have specifically targeted alow percentage of registered doctors, with no indication ofthe numbers who were not contactable, or who declined toparticipate in the survey [21–24].

There is also the possible assumption that more peopleinterested in the Internet might respond than non-interesteddoctors. If this is true, then it would also be true of the othersurveys to which this survey is compared. In addition, the datafrom the follow-up with the non-responders is statisticallyconsistent with the greater sample, therefore, the hypothesesthat the respondents were disproportionately Internet usersis highly unlikely.

In this survey, compared to the numbers in the SAMAdatabase, there was also the low response rate from theyounger groups. This was unexpected, because surveys fre-

quently have a disproportionally higher response from youngerGPs [25].

If one examines SA’s migration patterns, however, itbecomes clear that this is part of a much larger and well-

i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786

documented problem: the loss of doctors from South Africato developed countries. These losses are usually amongstthe younger doctors, many of whom view their qualifica-tion as their ticket to a better life elsewhere, and to pay offtheir study loans [26–33]. Although replacement recruitmentis ongoing, it is only marginally successful; partially becauseemigration from SA is officially under-reported, and SA doc-tors remain registered with local professional councils, usinglocal addresses [28,34].

The most important implication of the low response fromthe younger doctors might be that this survey under-reportsrather than over-reports the Internet usage rates amongst SAGPs.

Finally, the SAMA database itself appears to be out of date.Indications of this include the fact that only 14 of the 50 non-responders could be contacted, and that more than 700 GPs(4.1%) in the database are over the age of 65 (the oldest is 109).

In spite of the low response, because there has not yetbeen a study of SA GPs, and the fact that the demographicsof this sample so closely correspond to the demographics ofthe SAMA database, the results of this study are still worthusing.

4.4. Future directions

The three questions raised in the introduction to this paperhave been answered. In the process, however, further issueshave been raised for future investigation. There are:

1. 75% of GPs said that the Internet improves or significantlyimproves their practice of medicine. Just how has it doneso?

2. More than 80% of doctors said that their place of work hasaccess. Yet, only 72% of those who access the Internet do sofrom clinical practice or hospital, while 82% of those accessit from home. This difference is particularly pronounced inthe figures of female GPs. While the barriers are possiblypoor method of access and problems of time, this is not acertainty.

3. Contrary to expectations, GPs over 50 years access the Inter-net more frequently, although not necessarily for longer,than the others. Why is that?

4. To what extent are different usage patterns between PGPs,CGPs and HGPs reflective of differences in the health careorganisations, and to what extent do these usage patternsimpact on the health care delivered by those organisations?

5. How does the South African patient’s lack of Internet accessimpact upon the delivery of health care?

6. There is a strong suggestion that DoI’s “homophilous”social system is a major factor in the comparable usageof the Internet by SA GPs. This should be explored in moredetail.

5. Conclusion

In order to fill a gap in our understanding of the global use ofthe Internet by GPs, this paper details some of the results of thefirst national study of South African GPs’ use of the Internet.While the low response rate is of concern, given the demo-

i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n

Summary points

What was known before this study?

• Doctors’ Internet usage patterns in developed coun-tries, especially in the Northern Hemisphere, has beenwidely studied.

• No study of usage patterns amongst doctors in devel-oping countries exists.

• Therefore, a truly global picture of usage pattern is notcomplete.

What has this study added to the body of knowledge?

• Results from the first national survey of South AfricanGeneral Practitioners’ use of the Internet.

• In spite of South Africa’s being a developing coun-try, patterns closely resemble those of the developedworld, matching theoretical predictions of DoI.

• Further study is required to understand how the use ofthe Internet assists in the delivery of health care, andalso why home usage is disproportionally high.

gw

aaod

titS

C

Te

A

I

•

•

•••

r

College of Physicians.[21] American Medical Association [AMA], 2003 Physician

raphic correlations, it is unlikely that a higher response rateould have yielded markedly different figures.

While there are further areas of research, this paper hasrgued that, in spite of SA’s being a developing country, its GPs’doption and usage of the Internet is predicted by Diffusionf Innovations, and is equivalent to the usage patterns in theeveloped world.

This is significantly important for South Africa. Givenhe role of the Internet in the delivery of health care, asnternational medical practice increasingly embraces theseechnologies in the delivery of health care, indications are thatouth Africa will be positioned to play its role.

onflict of interest

he author has no conflict of interest. The survey was fundedntirely by personal funds.

cknowledgements

am indebted to

Dr Dick Ng’ambi and Prof Gail Todd, UCT, for comments onthis research.Marylin Myberg, SAMA, for supplying demographic andother information.Rauf Sayed, UCT, for guidance on the statistical analyses.

Members of UCT’s Division of Family Medicine journal club.UCT for funds for a notebook computer.

f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786 785

e f e r e n c e s

[1] S.S. Goudar, P. Kotur, Trends in Medical Education, IndianJournal of Anaesthesia 47 (2003) 25–29.

[2] R. Harden, J. Laidlaw, E. Hesketh, AMEE Medical EducationGuide No 16: study guides—their use and preparation,Medical Teacher 21 (1999) 248–265.

[3] M. Reddy, P. Dourish, A finger on the pulse: temporalrhythms and information seeking in medical work.Proceedings of CSCW02, ACM, New York, 2002.

[4] J.F. Burnham, M. Perry, Promotion of health informationaccess via Grateful Med and Loansom Doc: why isn’t itworking? Bulletin of the Medical Library Association 84(1996) 498–506.

[5] S.L. Leong, D. Gingrich, Lewis, R. Peter, D. Mauger, J.H.George, Enhancing doctor-patient communication usingemail: a pilot study, JABFP 18 (2005) 180–188.

[6] M.Y. Mann, M.A. Lloyd-Puryear, D. Linzer, Enhancingcommunication in the 21st century, Pediatrics 117 (2006)S315–S319.

[7] K. Masters, For what purpose and reasons do doctors usethe Internet: a systematic review, International Journal ofMedical Informatics 77 (2008) 5–16.

[8] E.M. Rogers, K.L. Scott, The diffusion of innovations modeland outreach from the national network of libraries ofmedicine to native American communities, http://nnlm.gov/pnr/eval/rogers.html.

[9] E. Rogers, Diffusion of Innovations, fifth ed., Free Press, NewYork, 2003.

[10] J. Eberhart-Phillips, K. Hall, G. Herbison, S. Jenkins, J.Lambert, R. Ng, M. Nicholson, L. Rankin, Internet useamongst New Zealand general practitioners, New ZealandMedical Journal 113 (2000) 135–137.

[11] International Telecommunication Union, Measuring theInformation Society: ICT Opportunity Index and WorldTelecommunication/ICT Indicators, ITU, Geneva,Switzerland, 2007.

[12] Economist Intelligence Unit, The 2006 e-readiness rankings,The Economist, London, 2006.

[13] National Department of Health, [of South Africa], RuralHealth Strategy - draft 2 March 2006, http://www.rudasa.org.za/download/RuralHealthStrat draft020306.doc.

[14] Statistics South Africa, Census 2001: Primary tables SouthAfrica Census’ 96 and 2001 compared, Statistics SouthAfrica, Pretoria, South Africa, 2004.

[15] A.S. Gosling, J.I. Westbrook, E.W. Coiera, Variation in the useof online clinical evidence: a qualitative analysis,International Journal of Medical Informatics 69 (2003) 1–16.

[16] J.I. Westbrook, A.S. Gosling, E. Coiera, Do clinicians useonline evidence to support patient care? A study of 55,000clinicians, Journal of the American Medical InformaticsAssociation 11 (2004) 113–120.

[17] S. Barclay, C. Todd, L. Finlay, G. Grande, P. Wyatt, Not anotherquestionnaire! Maximizing the response rate, predictingnon-response and assessing non-response bias in postalquestionnaire studies of GPs, Family Practice 19 (2002)105–111.

[18] B.R. McAvoy, General practice postal surveys: aquestionnaire too far? British Medical Journal 313 (1996)732–733.

[19] Canadian Medical Association [CMA].[20] D. Lacher, E. Nelson, W. Bylsma, R. Spena, Computer use and

needs of internists: a survey of members of the American

Resource Questionnaire Results, CMA, 2003, http://www.cmaj.ca/misc/prqindex.shtml.

i c a l

786 i n t e r n a t i o n a l j o u r n a l o f m e d

[22] P. Flanagan, S. Guy, C. Larsson, Staussois, EuropeanPhysicians and the Internet, Boston Consulting Group, 2003.

[23] R. Given, J. Girzadas, J. Bigalke, L. Meiterman, Taking thePulse: Physicians and Emerging Information Technologies,Deloitte Research and Fulcrum Analytics, 2002.

[24] C. van Knoop, D. Lovich, M. Silverstein, M. Tutty, Vitalsigns—E-health in the United States, Boston ConsultingGroup, 2003.

[25] L. Templeton, A. Deehan, C. Taylor, C. Drummond, J. Strang,Surveying general practitioners: does a low response ratematter? British Journal of General Practice 47 (1997) 91–94.

[26] P. Sullivan, Canada a prime destination as MDs flee SouthAfrica, Canadian Medical Association Journal 160 (1999)1615–1616.

[27] Padarath, C. Chamberlain, D. McCoy, A. Ntuli, M. Rowson, R.Loewenson, Health Personnel in Southern Africa:Confronting maldistribution and brain drain, (Loewenson, R;

Thompson, C (Eds.)), n.p. Regional Network for Equity inHealth in Southern Africa, Health Systems Trust (SouthAfrica) and MEDACT (UK), n.d.

[28] L.M. Nyhus, Y. Idezuki, D. Sheldon, I. McCusker, W.O. Griffen,Creation, evaluation and continuing assurance of excellence

i n f o r m a t i c s 7 7 ( 2 0 0 8 ) 778–786

of the certified surgical specialist, World Journal of Surgery24 (2000) 1519–1525.

[29] A. Adepoju, Continuity and changing configurations ofmigration to and from the Republic of South Africa,International Migration 41 (2003) 3–28.

[30] J. Boffa, Is there a doctor in the house? Australian and NewZealand Journal of Public Health 26 (2002) 301–304.

[31] J. Crush, The global raiders: nationalism, globalization, andthe South African brain drain, Journal of InternationalAffairs 56 (2002) 147–172.

[32] J. Eastwood, R. Conroy, S. Naicker, P. West, R. Tutt, J.Plange-Rhule, Loss of health professionals fromsub-Saharan Africa: the pivotal role of the UK, Lancet 365(2005) 1893–1900.

[33] R. Mattes, N. Mniki, Restless minds: South African studentsand the brain drain, in: Crush, Jonathan (Eds.), Idasa and

Southern African Research Centre, Queen’s University, CapeTown, South Africa, 2005.

[34] M. Awases, A. Gbary, J. Nyoni, R. Chatora, Migration ofHealth Professionals in Six Countries: a Synthesis Report,WHO Regional Office for Africa, n.p., 2004.