The economic consequences of social phobia

Journal of Affective Disorders 68 (2002) 221–233www.elsevier.com/ locate / jad

Research reportqThe economic consequences of social phobia

a , a,b a c*Anita Patel , Martin Knapp , Juliet Henderson , David BaldwinaCentre for the Economics of Mental Health, Institute of Psychiatry, 7 Windsor Walk, London SE5 8BB, UK

bPersonal Social Services Research Unit, London School of Economics, London, UKcMental Health Group, University of Southampton, Southampton, UK

Received 6 April 2000; received in revised form 26 September 2000; accepted 27 September 2000

Abstract

Background. Social phobia has been under-recognised and under-treated in many countries. Little is known about itseconomic impact. This study aimed to identify the economic consequences of social phobia for individuals, health servicesand wider society. Methods. Secondary analysis of 1993–1994 Psychiatric Morbidity Survey data compared 63 people withsocial phobia and 8501 people without psychiatric morbidity. Results. People with social phobia were less likely to be in thehighest socio-economic group and had lower employment rates and household income compared to those with no psychiatricmorbidity. They also had higher levels of drug dependency and use of prescribed oral medications. Although there were nodifferences in total health care costs, costs of GP contacts were significantly higher. Individuals with a comorbid psychiatricdisorder made higher use of some health services than those without a comorbidity. Limitations. Analyses were performedpost hoc on data collected for other purposes. The defining questions for social phobia have not been studied much before.The number of identified subjects is small and thus raises the possibility of type II errors. Larger numbers may have revealedeven more differences from the psychiatrically well population. Data on treatment patterns of the psychiatrically wellpopulation were limited because the surveys focused on subjects with psychiatric morbidity. Conclusions. The burden ofsocial phobia on individuals, health services and the wider society could be reduced through improved rates of detection andappropriate treatment. 2002 Elsevier Science B.V. All rights reserved.

Keywords: Social phobia; Service use; Costs; Economics

1. Introduction the disorder, its low rate of recognition and treatment(Jackson, 1992; Ross, 1991; Stein, 1996; Den Boer,

There is a need for better economic information on 1997), and the need for a baseline from which tosocial phobia stemming from the unknown burden of assess the consequences of successful treatment. Due

to ever present resource pressures, it is necessary toq know the costs of different ways of using scarceDeclaration of interest: This study was supported financially

resources, as well as outcomes and other conse-by SmithKline Beecham Pharmaceuticals.*Corresponding author. quences of intervention. The costs of social phobia

0165-0327/02/$ – see front matter 2002 Elsevier Science B.V. All rights reserved.PI I : S0165-0327( 00 )00323-2

222 A. Patel et al. / Journal of Affective Disorders 68 (2002) 221 –233

are of relevance to individual patients and their The Postcode Address File was used by OPCS asfamilies, and to primary care and specialist mental the sampling frame for the private households sur-health services. Data from other countries suggest veys. Of 18 000 addresses selected across UK,that social phobia imposes a substantial burden on Scotland and Wales, 15 765 were occupied by pri-health services (Schneier et al., 1992) and restricts vate households. From these, 12 730 eligible adultsearning potential (Wittchen and Beloch, 1996). This (aged 16–64) were selected for interview, 10 108study aimed to examine people with social phobia adults co-operated with the surveys, 1641 refusedliving in private households, identify impacts and and 981 were not contacted. Each sampled adult wasestimate the associated costs. Where possible, com- given Schedule A which covered socio-demographicparisons were made with people without psychiatric characteristics, general health questions, the Clinicalmorbidity. Interview Schedule–Revised (CIS–R) (Lewis and

Pelosi, 1990) and the Psychosis Screening Ques-tionnaire (Bebbington and Nayani, 1995). Those

2. Methods who scored at or above the threshold of 12 on theCIS–R or screened positive for psychosis were then

According to ICD-10 (World Health Organisation, given Schedule B. This covered long-standing ill-1993), social phobia is characterised by either: (1) ness, medication and treatment, use of health, socialmarked fear of being the focus of attention or fear of and voluntary care services, activities of daily living,behaving in a way that will be embarrassing or recent stressful life events, social activities, networkshumiliating; or (2) marked avoidance of being the and support, education, employment, finances, smok-focus of attention, or of situations in which there is a ing and alcohol consumption. Those below thefear of behaving in an embarrassing or humiliating threshold received Schedule C, an abridged versionway. of Schedule B, in order to obtain comparable in-

This marked fear and avoidance of scrutiny could formation on topics where there were no publisheddisadvantage people with social phobia at school, at data. 350 people with no psychiatric disorder com-work, and in forming and maintaining relationships. pleted Schedule B because they were falselySocial phobia might therefore be associated with: screened positive for psychiatric morbidity.lower educational attainment; greater likelihood ofeconomic inactivity and unemployment; higher 2.2. Identification of people with social phobialevels of absenteeism from work; lower socio-econ-omic status; higher levels of financial dependence The OPCS surveys applied algorithms to the CIS–(on others and on the state); higher use of health R scores to obtain (1 week) point prevalence ICD-10services; and higher direct and indirect costs. This diagnoses (World Health Organisation, 1993). Psy-study examined these possible consequences. chiatric disorders were then organised hierarchically

in order to allocate subjects with more than one2.1. The Psychiatric Morbidity Surveys disorder to just one diagnostic category. Social

phobia took priority over only one other disorder,Secondary analyses were carried out on data namely generalised anxiety disorder. This study

collected in the 1994–1995 Surveys of Psychiatric reports on all individuals diagnosed with socialMorbidity in Great Britain, conducted by the Office phobia, regardless of the hierarchical disorder as-of Population Censuses and Surveys (OPCS) (OPCS, signed to them. Published survey reports have only1996). Three separate surveys (each drawing upon a presented social phobia cases under the broadernational sample) were completed for private house- category of ‘all phobias’ (i.e. collapsed withholds, institutions (including hospitals and residential agoraphobia and specific phobia) (Meltzer et al.,homes) and homeless people. Here we focus on the 1995).private households sample because that is where We used this same algorithm to identify an ICD-most people with social phobia are likely to be 10 diagnosis of social phobia: (a) fear and scrutinyfound. by other people: fear of eating or speaking in public;

A. Patel et al. / Journal of Affective Disorders 68 (2002) 221 –233 223

(b) social impairment; (c) avoidant behaviour a employed or economically inactive, the overall lostprominent feature; and (d) overall phobia score of employment costs will anyway be an underestimate.$ 2 in the CIS–R.

2.6. Costs

2.3. Alcohol and drug useAll costs were calculated in pounds sterling at

1997–1998 price levels. Service cost estimates wereAll subjects received a self-completion ques-based on figures in Netten et al. (1998). These aretionnaire about alcohol and drug dependence. Al-well-based, national estimates of long-run marginalcohol dependence was measured using 12 questionscosts and include allowances for capital, overheads,covering three components: loss of control, symp-travelling time, non-patient contact time and supporttomatic behaviour and binge drinking (Clark andservices (such as clerical support). National averageHilton, 1991). A minimal problem was indicated bywage rates (New Earnings Survey, 1997) wereone to two affirmed items, a moderate problem byattached to days taken off work by those in employ-three items and a high level problem by $ 4 items.ment (1996–1997 figures were inflated to 1997–Drug dependence was measured using five questions1998 prices). All social security benefits were valuedabout frequency of use, stated dependence, inabilityusing 1998 payment rates (Benefits Agency, 1998).to cut down, need for larger amounts and withdrawal

Costs were not attached to two additional items ofsymptoms (Robins and Regier, 1991). A positiveresource use data. Firstly, costs could not be esti-response to at least one question indicated somemated for contacts with professionals for drug-re-dependence.lated problems because the surveys did not ask aboutthe type of professional seen, or the number of

2.4. Service utilisation contacts. Secondly, costs were not applied to receiptof medication in order to avoid double-counting

The surveys asked about inpatient stays, outpatient prescription elements already included in serviceepisodes, counselling or therapy contacts, home costs.visits from health or social services or voluntaryworkers in the previous twelve months; and about 2.7. Adjustments to dataGP consultations in the previous 2 weeks. Tenpeople with social phobia did not have full service Some adjustments were necessary to obtain fulluse data because they had not completed Schedule B. service costs. These were applied across the fullIn terms of the psychiatrically well population, only sample of 10 108 in order to avoid complex multiplethose 350 false positive people who mistakenly adjustments for those with comorbidities.received Schedule B had service use data.

2.7.1. GP consultations2.5. Employment and benefits It was assumed that survey responses about GP

consultations in the previous 2 weeks could beA large part of Schedule B was dedicated to extrapolated to an annual cost. 39 out of 1617

economic activity in terms of current employment subjects with a GP consultation but no data onstatus, employment history of individuals who were contact frequency were allocated the median cost fornot currently working, time taken off work, receipt those who did have data.of state benefits, individual income and householdincome. Costs were applied to time taken off work 2.7.2. Outpatient visitsdue to illness and to state benefits. As social security Despite detailed questions about the settings inbenefits are transfer payments, there is a danger of which outpatient visits took place, there were sub-double counting other costs e.g. some of the lost stantial missing data. Consequently, costs for aemployment costs. However, as lost employment generic NHS hospital outpatient department werecosts were not estimated for those who were un- used. 756 out of 863 people with an outpatient visit


had no data on contact frequency. These were (‘social phobia with comorbidity’), and those withallocated the median cost for those who did provide no identified psychiatric disorders (‘psychiatricallysuch data. well population’). The ‘social phobia only’ and

‘social phobia with comorbidity’ groups are sub-groups of ‘all social phobia’. The all social phobia

2.7.3. Home visits group was compared with the psychiatrically wellResponses to questions about frequency of home population using ANOVA with linear contrasts in

visits were in the following categories: four or more order to allow for any differences between the twotimes a week (for costing purposes assumed to be social phobia subgroups. The two subgroups wereequivalent to 208 times per annum); two or three compared against each other and against thetimes a week (104 per annum); once a week (52 per psychiatrically well population using the Student’sannum); at least once a month (24 per annum); and t-test. A statistical significance threshold of P , 0.05less than once a month (8 per annum). 13 out of 142 was used. Only significant test results are reported inpeople who received home visits but had no data on the tables, except in the case of Table 8.frequency of visits or type of professional seen were Published survey reports were based on weightedallocated the median annual home visit cost (based data to allow for non-response associated withon sum of all professionals) for those who did have household size, different probabilities of selectingsuch data. subjects in different sized households and to repre-

sent the age–sex structure of the total nationalpopulation living in private households. These

2.7.4. Counselling /therapy weights were left in for the analyses reported hereAlthough questions about type of contact and (with the exception of the reporting of demographic

setting were detailed, it was not possible to cost characteristics) and gave rise to minor data ir-these individually due to large amounts of missing regularities.data. Therefore, a NHS hospital-based clinical psy-chologist visit was assumed. 3 out of 162 peoplewho received counselling / therapy but had no fre-quency data were allocated the median annual coun- 3. Resultsselling / therapy cost for those with data.

3.1. Prevalence

2.7.5. State benefits Sixty-three people met the criteria for socialMaternity allowance, child benefit and war phobia, representing a point prevalence rate of 0.6%

widows pension were not included as they are clearly among the population of adults aged 16–64 residentunrelated to mental illness. As questions relating to in private households. There were 8501 subjects withthe receipt of benefits referred to the previous week, no psychiatric disorders.it was necessary to extrapolate costs to a 1-yearperiod.

3.2. Demographic characteristics

2.8. Analyses There were 31 men and 34 women with socialphobia. Eighteen (28%) were aged under 25 years.

Differences were tested across four groups: all Ninety-one-percent of people with social phobiathose who fulfilled the criteria for social phobia (‘all were white, 3% were Pakistani, 2% were blacksocial phobia’), those who fulfilled the criteria for Caribbean, 2% were black other and 3% weresocial phobia and no other psychiatric disorders categorised as ‘other’. Thirty-one-percent were mar-(‘social phobia only’), those who fulfilled criteria for ried, 12% cohabiting, 37% single, 2% widowed, 6%social phobia and comorbid psychiatric disorders divorced and 12% separated.


3.3. Comorbidity the social phobia only group compared to thecomorbid group.

Twenty-seven people had a comorbid psychiatricdisorder: depression (five severe, five moderate, two 3.5. Education, economic status and employmentmild), obsessive compulsive disorder (nine) andpsychosis (two). Of these 27 people, 13 also had a All four groups had similar rates of people attain-long-standing physical illness. Physical illness was ing the highest qualification and attaining no qualifi-also present in nine of the 36 people with no other cations (Table 2). However, compared to thepsychiatric disorder than social phobia. psychiatrically well population, the all social phobia

group were less likely to have a teaching/nursingqualification or HND (F 5 1.660, P 5 0.190;2,8533

3.4. Alcohol and drug dependence value of contrast 5 0.073, df 5 55.91, P , 0.01).Socio-economic group was rated by the OPCS on

The prevalence of alcohol dependence was 20.4% the basis of the subject’s own most recent occupa-for all social phobia, and 14.7% for the psychiatrical- tion, except for married or cohabiting women forly well population (Table 1). There were no signifi- whom a spouse’s or partner’s occupation was usedcant differences between the four groups. However, (the woman’s own occupation was only used if herthe all social phobia group had significantly higher spouse or partner had never worked). Table 3 showslevels of drug dependency (F 5 53.49, P , that there were fewer people with social phobia2,8561

0.0001; value of contrast 5 2 0.11, df 5 51.89, P , (1.6%) in the top category of professional, compared0.01) and drug problems (F 5 41.95, P , to the psychiatrically well population (7.1%)2,8561

0.0001; value of contrast 5 2 0.069, df 5 34.84, P , (F 5 1.529, P 5 0.217; value of contrast 52,8561

0.05) compared to the psychiatrically well popula- 0.057, df 5 37.56, P , 0.0001). This is probably duetion. This was mainly due to higher levels of to the comorbid group who had fewer people in thedependency (P , 0.05) and problems (P , 0.05) in professional category than the psychiatrically well

Table 1Alcohol and drug dependence

(%) Social Social phobia All Psychiatricallyphobia with social wellonly comorbidity phobia population

Alcohol dependenceNo problem 87. 5 69.3 79.6 85.3Minimal 6.4 8.6 7.4 10.6Moderate 1.5 10.4 5.3 2.0High 4.6 11.7 7.7 2.1n 36 27 63 8184

Drug dependenceNo dependence 78.4 96.2 86.1 98.7

a,b,cDependence 21.6 3.8 13.9 1.3No problem (%) 84.5 100.0 91.2 99.1

d,e,fProblem 15.5 – 8.8 0.9n 36 27 63 8501

a All social phobia vs. well population: F 5 53.49, P , 0.0001; value of contrast 5 2 0.11, df 5 51.89, P , 0.01.2,8561b Social phobia only vs. social phobia with comorbidity significant at P , 0.05 level.c Social phobia only vs. well population significant at P , 0.01 level.d All social phobia vs. well population: F 5 41.95, P , 0.0001; value of contrast 5 2 0.069, df 5 34.84, P , 0.05.2,8561e Social phobia only vs. social phobia with comorbidity significant at P , 0.05 level.f Social phobia only vs. well population significant at P , 0.05 level.


Table 2Highest educational attainment

Qualification level Social Social All Psychiatrically(in descending order phobia phobia with social wellof attainment) (%) only comorbidity phobia population

(n 5 36) (n 5 27) (n 5 63) (n 5 8474)

Degree 11.9 12.1 12.0 11.4

Teaching qualification /Nursingqualification /Higher National

a,bDiploma 7.9 1.9 5.3 12.2

Advanced Level 7.0 9.7 8.2 10.9

Ordinary Level 37.7 40.1 38.7 25.7

Certificate of SecondaryEducation 12.1 15.0 13.4 10.6

Certificate of SecondaryEducation no grade – – – 0.2

Other – – – 0.8

No qualification 23.4 21.2 22.5 27.4a All social phobia vs. well population: F 5 1.660, P 5 0.190; value of contrast 5 0.073, df 5 55.91, P , 0.01.2,8533b Social phobia with comorbidity vs. well population significant at P , 0.01 level.

population (P , 0.0001). Removing women who People with social phobia had much lower rates ofwere married /cohabiting from the analyses (in order employment than the psychiatrically well populationto only include those individuals whose own occupa- (Table 4). For example, 31% of the all social phobiation was recorded) did not alter the findings. group were working full-time compared to 54% of

the psychiatrically well population (F 5 7.279,2,8546

Table 3 P , 0.01; value of contrast 5 0.24, df 5 59.46, P ,Socioeconomic status 0.0001). These differences persisted when separatelySocioeconomic group All Psychiatrically comparing people with and without comorbidities(%) social well with the psychiatrically well population. These find-

phobia population ings are reflected in the rates of unemployment and(n 5 63) (n 5 8501)

economic inactivity, with the all social phobia groupa,bProfessional 1.6 7.1 being more likely to be unemployed (F 5 6.032,2,8546

Employers and managers 17.2 19.6 P , 0.01; value of contrast 5 2 0.12, df 5 56.23,Intermediate non-manual 8.4 10.8

P , 0.05) and economically inactive (F 52,8546Junior non-manual 18.9 10.97.732, P , 0.0001; value of contrast 5 2 0.20, df 5Skilled manual 26.2 28.8

Semi-skilled manual 14.8 14.4 55.47, P , 0.01). This difference was not presentUnskilled manual 6.1 4.6 when separately comparing the social phobia onlyArmed forces – 1.1 group with the psychiatrically well population, butNever worked 6.1 2.6

was when comparing the comorbid group (P , 0.01).Insufficient information – 0.2Twenty-four-percent of the all social phobiaa All social phobia vs. well population: F 5 1.529, P 52,8561 group, compared to only 5% in the psychiatrically0.217; value of contrast 5 0.057, df 5 37.56, P , 0.0001.

b well population, reported leaving a job in the lastSocial phobia with comorbidity vs. well population significantat P , 0.001 level. year due to mental, nervous or emotional problems


Table 4Employment status and number of days taken off work in the past year

Social Social All Psychiatricallyphobia phobia with social wellonly comorbidity phobia population

Employment statusa,b,cWorking full-time (%) 33.4 26.7 30.5 54.3d,eWorking part-time (%) 14.8 3.6 9.9 16.6

fUnemployed (%) 18.5 20.0 19.2 7.5g,hEconomically inactive (%) 33.3 49.8 40.4 21.6

n 36 27 63 8486

iNumber of days taken off work in past year0 (%) 41.5 69.1 53.9 65.81–3 (%) 5.8 12.6 8.9 7.84–6 (%) 10.3 – 5.7 8.67–13 (%) 10.5 – 5.8 7.214–20 (%) 22.6 – 12.4 3.821–27 (%) 9.3 18.3 13.4 0.728–55 (%) – – – 2.756–91 (%) – – – 1.192–182 (%) – – – 1.1365 (%) – – – 1.2n 10 8 18 222

a All social phobia vs. well population: F 5 7.279, P , 0.01; value of contrast 5 0.24, df 5 59.46, P , 0.0001.2,8546b Social phobia only vs. well population significant at P , 0.05 level.c Social phobia with comorbidity vs. well population significant at P , 0.01 level.d All social phobia vs. well population: F 5 1.703, P 5 0.182; value of contrast 5 0.074, df 5 56.43, P , 0.05.2,8546e Social phobia with comorbidity vs. well population significant at P , 0.01 level.f All social phobia vs. well population: F 5 6.032, P , 0.01; value of contrast 5 2 0.12, df 5 56.23, P , 0.05.2,8546g All social phobia vs. well population: F 5 7.732, P , 0.0001; value of contrast 5 2 0.20, df 5 55.47, P , 0.01.2,8546h Social phobia with comorbidity vs. well population significant at P , 0.01 level.i Due to physical and mental ill-health and only for those people in employment at time of survey.

(F 5 7.513, P , 0.01; value of contrast 5 3.357, P , 0.05; value of contrast 5 2 0.208, df 52,111

2 0.17, df 5 35.35, P , 0.05). There were no statisti- 268, P , 0.05). However, there were no differencescally significant differences between the two social in the number of people in the top band of individualphobia subgroups although a smaller proportion of or household income.the social phobia only group (10%) left their job forthese reasons compared to the comorbid group 3.6. Service utilisation(34%). For those currently employed, there were nodifferences in the proportion that had taken time off Across the all social phobia group, service use waswork in the past year due to ill health (mental and modest (Table 6). Examination of service use pat-physical). However, it can be noted that the rates for terns showed that only 71% of the all social phobiaall groups were generally high at 46, 59, 31 and 34% group had used at least one service compared to 91%for the all social phobia group, social phobia only of the psychiatrically well population (F 521 468

group, social phobia with comorbidity group and 13.016, P , 0.0001; value of contrast 5 0.20, df 5

psychiatrically well population respectively. 49.71, P , 0.01). Those with social phobia onlyPeople with social phobia were more likely to were less likely to have used a service than those

have a low ( # £200) household income than the with a comorbidity (63% vs. 79%) although thispsychiatrically well population (Table 5; F 5 difference was not significant. Although there were2,268


Table 5Individual and household gross incomes

aWeekly income Individual weekly gross income Household weekly gross income(£, 1993/94)(%) All social Psychiatrically All social Psychiatrically

phobia well population phobia well population(n 5 53) (n 5 347) (n 5 51) (n 5 332)

0.01–100 56.7 34.9 14.9 7.2100.01–200 18.5 25.6 25.5 17.3200.01–300 10.3 14.2 11.2 11.4300.01–400 2.3 7.2 2.4 11.0400.01–500 2.8 4.1 1.0 7.0500.01 or more 5.9 4.9 16.6 16.9NA/DK 3.6 9.0 28.4 29.2

a Proportion of people with household income # £200: F 5 3.357, P , 0.05; value of contrast 5 2 0.208, df 5 268, P , 0.05.2,268

no individuals in any group who had used all five the psychiatrically well population (1%) (F 52,1468

services, a higher proportion of the all social phobia 54.39, P , 0.0001; value of contrast 5 2 0.16, df 5

group (17%) had used three or more services than in 45.62, P , 0.01). When comparing the use of in-

Table 6Percentage of people using health services and mean number of contacts

Service users and mean use Social Social All Psychiatricallyphobia phobia social wellonly comorbidity phobia population

Inpatient services in the past yeard,e% 1.8 20.6 11.7 12.0

aMean days 0.1 1.0 0.6 1.6

bOutpatient episodes in the past year% 53.1 61.7 57.6 43.0

aMean number 1.8 2.6 2.3 1.8

Home visits by health and social services in the past yearf% 2.1 19.5 11.3 5.6

aMean number 1.1 6.5 3.9 1.4

Counselling /therapy contacts in the past year% 6.6 13.0 10.0 6.8

aMean number 3.2 0.5 1.8 0.6n 25 27 52 340

General practitioner contacts in the past 2 weeksg,h% 19.0 37.1 26.8 14.0

cMean number 1.3 1.1 1.2 1.2n 36 27 63 8185

a Mean utilisation over full valid sample, not just for users.b Includes outpatient department, A&E, health centre, private consultancy room.c Mean utilisation only for those who saw GP in past 2 weeks.d Social phobia only vs. social phobia with comorbidity significant at P , 0.05 level.e Social phobia only vs. well population significant at P , 0.01 level.f Social phobia only vs. social phobia with comorbidity significant at P , 0.05 level.g All social phobia vs. well population: F 5 6.330, P , 0.01; value of contrast 5 0.14, df 5 50.28, P , 0.05.2,8245h Social phobia with comorbidity vs. well population significant at P , 0.05 level.


dividual services, those with a comorbidity had nervous system medications, compared to 6.1% ofstatistically higher rates of some service use than the psychiatrically well population (F 5 8.325,2,8561

those without a comorbidity: 21% vs. 2% using P , 0.0001; value of contrast 5 2 0.12, df 5 52.49,inpatient services (P , 0.05) and 20% vs. 2% receiv- P , 0.05); and 9% used antidepressants, compared toing home visits (P , 0.05). Differences in the rates 0.2% of the psychiatrically well populationof GP use were not statistically different between (F 5 99.142, P , 0.0001; value of contrast 52,8561

these two sub-groups. However, more of the all 2 0.079, df 5 60.86, P , 0.05). There were no sig-social phobia group (27%) visited a GP than the nificant differences in the use of these two types ofpsychiatrically well population (14%) (F 5 medications between those with social phobia only2,8245

6.330, P , 0.01; value of contrast 5 0.14, df 5 50.28, and those with a comorbidity.P , 0.05).

3.8. Costs3.7. Medication

Table 8 summarises the annual health service andThe percentage of people taking oral medication lost employment costs and the value of benefits

was higher for the all social phobia group (44%) received. The mean cost of GP visits amongst the allthan for the psychiatrically well population (27%) social phobia group is almost twice as much as that(F 5 4.534, P , 0.05; value of contrast 5 0.17, for the psychiatrically well population (F 52,8244 2,8245

df 5 57.36, P , 0.05). There were also differences in 3.809, P , 0.05; value of contrast 5 2 64.23, df 5

the type of medications used (Table 7). For example, 55.01, P , 0.05). This difference is most apparent18% of the all social phobia group used central when comparing the social phobia with comorbidity

Table 7Use of medications

Use of medications Social phobia Social phobia All social Psychiatricallyonly with cormorbidity phobia well population

aTaking any medication by mouth (%) 43.9 44.7 44.3 27.3n 36 27 63 8184Taking regular course of injections (%) – 8.2 3.5 1.5n 36 27 63 8175

bTaking any CNS drugs (%) 15.2 21.7 18.0 6.1n 36 27 63 8501Breakdown of medications (%)Betablockers 2.4 5.6 3.8 0.1Nitrate and Ca-channel blockers – 3.6 1.6 0.1Barbiturates – – – –Hypnotics and Anxiolytics – 12.7 5.5 0.1Antipsychotic drugs – – – 0.3Antipsychotic depot – – – 0.0Antimanic – – – 0.1Drugs used in psychoses and related – – – 0.4

c,dAntidepressants 11.0 5.2 8.5 0.2CNS stimulants – – – 0.0Appetite suppressants – – – 0.0Appetite suppressants and bulkforming – – – 0.0Antimuscarinic – – – 0.0Drugs used in substance dependence – – – –n 36 27 63 8501

a All social phobia vs. well population: F 5 4.534, P , 0.05; value of contrast 5 0.17, df 5 57.36, P , 0.05.2,8244b All social phobia vs. well population: F 5 8.325, P , 0.0001; value of contrast 5 2 0.12, df 5 52.49, P , 0.05.2,8561c Social phobia only vs. well population significant at P , 0.05 level.d All social phobia vs. well population: F 5 99.142, P , 0.0001; value of contrast 5 2 0.079, df 5 60.86, P , 0.05.2,8561

230A

.P

atelet

al./

Journalof

Affective

Disorders

68(2002)

221–233

Table 8Annual costs of health services, lost employment and benefits (£, 1997/1998 prices)

b cSocial phobia only Social phobia with All social phobia Psychiatrically well P value ANOVA

comorbidity population

Mean (S.D.) Mean (S.D.) Mean (S.D.) Mean (S.D.) ANOVA CONTRAST TEST

F P value Contrast value, df, P valuedf

Health service costs

GP 96.10 (216.41) 164.58 (234.44) 125.78 (225.16) 66.12 (196.78) 0.234 3.809 0.022 2 64.23, 55, 0.0302,8245

n 36 27 63 8185

Inpatient 11.59 (86.64) 216.87 (684.54) 119.26 (505.58) 345.55 (2465.5) 0.132 0.269 0.765 231.32, 389, 0.5012,389

Outpatient 110.40 (121.00) 157.16 (146.59) 134.92 (135.79) 110.42 (173.93) 0.218 0.968 0.381 2 23.36, 389, 0.3552,389

Home visits 27.63 (191.50) 183.21 (456.46) 109.23 (361.33) 34.49 (173.26) 0.111 6.671 0.001 2 70.93, 39, 0.1522,389

Counselling 177.13 (680.77) 30.03 (98.84) 99.98 (475.57) 89.98 (770.30) 0.297 0.262 0.770 2 13.60, 389, 0.9022,389

n 25 27 52 340

Total health care costs 451.84 (821.25) 751.85 (927.77) 609.19 (883.25) 379.09 (969.69) 0.224 1.928 0.147 2 222.76, 367, 0.1212,367

n 25 27 52 318

Lost employment costsaCosts of days off work 538.49 (634.65) 320.92 (677.17) 440.81 (644.46) 594.76 (2245.2) 0.487 0.065 0.937 165.06, 237, 0.7562,237

n 10 8 18 222

Value of social security benefits

Benefit value 1106.22 (1506.32) 1816.52 (2422.45) 1478.75 (2051.1) 794.16 (1519.04) 0.215 5.439 0.005 2 667.21, 53, 0.0252,386

n 25 27 52 337

a Days taken off work due to ill health for those people in employment at time of survey.b Using t-test to compare social phobia only with social phobia with comorbidity.c Using one way ANOVA polynomial contrasts to compare well population, social phobia only and social phobia with comorbidity (coefficients 5 1, 2 0.5, 2 0.5

respectively).


group to the psychiatrically well population (P , well, it also implies that we may have underesti-0.05). Generally, those with a comorbidity had mated the extent to which the costs of social phobiahigher costs than those with social phobia only (with are higher than the well population. Despite thesethe exception of counselling and lost employment limitations, the surveys offer the broadest basis forcosts) although these were not statistically different. this kind of work. Consequently, this is the first time

Summing these direct and indirect costs provides a a UK study has been able to identify the economicconservative indication of the aggregate costs of burden of social phobia in the general population.social phobia because it excludes the costs of otherservices, the burden on families, the intangible lossesto quality of life of sufferers and mortality effects. 4.1. Prevalence

Point /1 month prevalence rates for other countriesvary between 1.3 in the USA (Boyd et al., 1990) to

4. Discussion 14.2 in Stockholm (Furmark et al., 1999), althoughFaravelli et al. (1989) reported a rate of 0.45 for

Anxiety disorders have been estimated to cost Florence. Lifetime estimates range from 0.99 to 16%$46.6 billion annually in the US (DuPont et al., (Lecrubier et al., 2000). Obviously it is difficult to1996). However, there is a lack of information make comparisons without accounting for differingspecifically on social phobia, particularly in the UK. sampling procedures and diagnostic criteria (almostThis study presents new insights into the economic all other studies have used DSM criteria for diag-burdens of social phobia. As the Psychiatric Mor- nosis), but the 1 week prevalence rate of 0.6%bidity Surveys were cross-sectional, there were some reported here is generally lower than found else-limitations for the analyses. Exact service use pat- where. One possible explanation is provided in aterns could not be established due to missing data, recent cross validation study comparing the CIS–Rthe survey relied on the recall of people with with SCAN ICD-10 diagnostic categories whichdisorders, and approximate unit costs had to be used found poor agreement between these two schedulesfor some services. Additionally, the number of in the identification of a large range of ICD-10identified subjects is small and thus raises the neurotic disorders (Brugha et al., 1999). The poorpossibility of type II errors. Larger numbers may sensitivity of the CIS–R for SCAN phobic disordershave revealed even more differences from the was notable, suggesting that the OPCS surveys maypsychiatrically well population. Finally, data on have greatly underestimated the prevalence of thesetreatment patterns of the psychiatrically well popula- disorders.tion were limited because the surveys focused onsubjects with psychiatric morbidity. We further ack-nowledge that the use of the false positive cases to 4.2. Burden on individualsrepresent a well population in some of the analysesmay have limitations as they may have just fallen People with social phobia can experience impair-short of the threshold for presence of a psychiatric ments in their personal, academic and professionaldisorder. In order to check the representativeness of lives (Schneier et al., 1992; Wittchen and Beloch,this group, we compared their CIS–R scores with the 1996). This study shows that a higher proportion ofremainder of the well population. The false positive people with social phobia were single, divorced andcases had a mean CIS–R score of 4.81, which was separated than in the psychiatrically well population.significantly higher than the mean score of 2.93 for Leon et al. (1995) found high levels of financialthe remainder of the well population (P , 0.001). dependence and unemployment, which we have alsoThe proportion of false positive cases that had a found:score of ten or 11 (11%) was twice as high as for theremainder (5%). Although this suggests that the false • those with social phobia were more likely to bepositive cases were not completely psychiatrically economically inactive (not available for work);


• if economically active, to be currently unem- comorbid subgroup. Rates of antidepressant use wereployed; of interest, being twice as high in the social phobia

• and for the whole group to have lower household only group although depression was the most com-incomes. mon co-existing condition in the comorbid group.

Many people with social phobia use alcohol to4.3. Burden on health and social services reduce anxiety (Den Boer, 1997) but we found no

significant differences in levels of alcohol depen-US studies have shown that only a small propor- dence. However, our finding of a higher rate of drug

tion of individuals with social phobia in the com- dependence is consistent with reports of drug abusemunity seek professional help specifically for their being commonly associated with social phobiasocial phobia (Schneier et al., 1992; Magee et al., (Schneier et al., 1992).1996). Although they do consult frequently, it isoften the presence of a comorbid disorder that 4.4. Burden on othersprompts them to present to their GP. This study alsoshows that people with social phobia have a higher It is obvious that the consequences of socialrate of GP consultations than the psychiatrically well phobia are far-reaching and thus incur wider societalpopulation but there were no differences between costs, mainly due to economic inactivity, unemploy-social phobia sufferers with and without comor- ment, absenteeism from work and financial depen-bidities. dence on the state. The lost employment costs

In terms of other service use, Schneier et al. estimated here are only for days taken off work by(1992) found that individuals with pure social phobia those in employment. They do not include thoseattended medical outpatient clinics twice as frequent- arising from unemployment and economic inactivi-ly as those with no disorder. The presence of a ty–inclusion of costs for these groups is a contenti-comorbid condition increased this to a three-fold ous issue in economics as it could be argued that in adifference, with a further increase in the use of society without full employment, the vacated jobspsychiatric inpatients, psychology outpatients and could be filled by other unemployed people.emergency departments. Although our data did notallow further service use comparisons against the 4.5. Primary care implicationspsychiatrically well population, we did find thatsocial phobia sufferers with a comorbidity had higher There is growing evidence on the availability ofrates of inpatient service use and home visits than safe and effective psychotherapeutic and psycho-those with social phobia only. However, overall pharmacological treatments for social phobia (Leonservice use by the social phobia group was fairly et al., 1995; Costa e Silva, 1998; Ballenger et al.,modest. This accords with Leon et al. (1995) who 1998; Nutt et al., 1999). However, under-recognitionfound that, although reported service use for all and consequent inappropriate treatment and health-phobias collectively was higher than for those with seeking behaviour continue to place large burdens onno disorder, it was considerably lower than for those health services. Comorbid psychiatric problems arewith panic disorder and obsessive compulsive disor- common among people with social phobia (Dender. Boer, 1997) — although social phobia usually

Nearly half of the social phobia group were taking precedes them (Schneier et al., 1992; Lecrubier,prescribed oral medications, which is significantly 1998). They play an important role because theyhigher than the rate of 27% in the psychiatrically often prompt sufferers to seek healthcare. However,well population. However, if it is the case that social while secondary disorders may be recognised andphobia is treatable pharmacologically, for example treated, social phobia may remain undetected. Costawith SSRIs (Ballenger et al., 1998), this would imply e Silva (1998) urged large health organisations tothat over half of the sample were not being treated make concerted efforts to bring about changes to thisappropriately. In fact, 41% of those on no medication pattern.had also not used any health services. Interestingly, As primary care remains the first port of call foroverall medication use was no more common in the the majority of people seeking healthcare, this ap-


disorders: a WHO perspective. Acta Psychiatr. Scand. 393pears to be the most obvious place to start. If we(Suppl.), 2–5.extrapolate our data to a Primary Care Group

Den Boer, J.A., 1997. Social phobia: epidemiology, recognitionpopulation of 100 000 people (based on 1991 UK and treatment. BMJ 27, 796–800.population census statistics on the number of adults DuPont, R.L., Rice, D.P., Miller, L.S. et al., 1996. Economic costsliving in private households), total healthcare costs of of anxiety disorders. Anxiety 2, 167–172.

Faravelli, C., Guerrini Degl’Innocenti, B., Giardinelli, L. et al.,social phobia would amount to over £195,000 per1989. Epidemiology of anxiety disorders in Florence. Actaannum, with primary care costs alone at £49,000.Psychiatr. Scand. 79, 308–312.

Wider costs such as social security benefit claims Furmark, T., Tillfors, M., Everz, P.-O. et al., 1999. Social phobiawould stand at £474,000. Educating primary care in the general population: prevalence and sociodemographicstaff about social phobia and its treatment would be profile. Soc. Psychiatr. Psychiatr. Epidemiol. 34, 416–424.

Jackson, S.W., 1992. The listening healer in the history ofan important step towards reducing the individualpsychological healing. Am. J. Psychiatr. 149, 1623–1632.and societal burden of this disorder.

Lecrubier, Y., 1998. Comorbidity in social anxiety disorder:impact on disease burden and management. J. Clin. Psychiatr.59 (Suppl. 17), 33–37.

Acknowledgements Lecrubier, Y., Wittchen, H.-U., Faravelli, C. et al., 2000. AEuropean perspective on social anxiety disorder. Eur.Psychiatr. 15, 5–16.We thank Professor Glyn Lewis for his assistance

Lewis, G., Pelosi, A.J., 1990. Manual of the Revised Clinicalwith identifying the sample from the survey data, Interview Schedule, (CIS–R). Institute of Psychiatry, London.Jack Astin for statistical support and Dr Julia Bot- Leon, A.C., Portera, L., Weissman, M.M., 1995. The social coststomley for helpful comments on the preparation of of anxiety disorders. Br. J. Psychiatr. 166 (Suppl. 27), 19–22.

Magee, W.J., Eaton, W.W., Wittchen, H.-U. et al., 1996.this manuscript. We also thank The Data Archive forAgoraphobia, simple phobia, and social phobia in the Nationalproviding the OPCS dataset. Crown copyright ma-Comorbidity Survey. Arch. Gen. Psychiatr. 53, 159–168.

terial reproduced by permission of the Controller of Meltzer, H., Gill, B., Petticrew, M. et al., 1995. Report 1: TheHer Majesty’s Stationary Office. The copyright Prevalence of Psychiatric Morbidity among Adults Living inholder, the original data producer, the relevant Private Households. OPCS Surveys of Psychiatric Morbidity in

Great Britain. OPCS, London.funding agencies and The Data Archive bear noNetten, A., Dennett, J., Knight, J., 1998. The Unit Costs of Healthresponsibility for their further analysis and interpre-

and Social Care. PSSRU. University of Kent at Canterbury.tation. New Earnings Survey, 1997. Part A: Streamlined and Summary

Analysis. HMSO, London.Nutt, D., Baldwin, D.S., Beaumont, G. et al., 1999. Guidelines for

the management of social phobia / social anxiety disorder.ReferencesPrimary Care Psychiatr. 5 (4), 147–155.

Office of Population Consuses and Surveys Social SurveysBallenger, J.C., Davidson, J.R.T., Lecrubier, Y. et al., 1998. Division, OPCS Surveys of Psychiatric Morbidity: Private

Consensus statement on social anxiety disorder from the Household Survey, 1993 (computer file). Colchester, Essex:International Consensus Group on Depression and Anxiety. J. The Data Archive (distributor), 24 September 1996. SN: 3560.Clin. Psychiatr. 59 (Suppl. 17), 54–60. Robins, L.N., Regier, D.A., 1991. Psychiatric Disorders in

Bebbington, P.E., Nayani, T., 1995. The Psychosis Screening America: the Epidemiological Catchment Area Study. The FreeQuestionnaire. Int. J. Meth. Psychiatr. Res. 5, 11–20. Press (Macmillan Inc), New York.

Benefits Agency, 1998. Social Security Benefit Rates from April Ross, J., 1991. Social phobia: the Anxiety Disorders Association1998. of America helps raise the veil of ignorance. J. Clin. Psychiatr.

Brugha, T.S., Bebbington, P.E., Jenkins, R. et al., 1999. Cross 52 (Suppl 11), 43–47.validation of a general population survey diagnostic interview: Schneier, F.R., Johnson, J., Hornig, C.D. et al., 1992. Sociala comparison of CIS–R with SCAN ICD-10 diagnostic phobia: comorbidity and morbidity in an epidemiologicalcategories. Psychological Medicine 29, 1029–1042. sample. Arch. Gen. Psychiatr. 49, 282–288.

Boyd, J.H., Rae, D.S., Thompson, J.W. et al., 1990. Phobia: Stein, M.B., 1996. How shy is too shy? Lancet 347, 1131–1132.prevalence and risk factors. Soc. Psychiatr. Epidemiol. 25, Wittchen, H.U., Beloch, E., 1996. The impact of social phobia on314–323. quality of life. Int. Clin. Psychopharmacol. 11 (Suppl. 3),

Clark, W.B., Hilton, M.E. (Eds.), 1991. Alcohol in America: S15–S23.Drinking Practices and Problems. State University of New World Health Organisation, 1993. The ICD-10 Classification ofYork Press, Albany. Mental and Behavioural Disorders: Diagnostic Criteria for

Costa e Silva, J.A., 1998. The public health impact of anxiety Research. WHO, Geneva.

Documents

The economic consequences of social phobia