1 Simms I, Rogers PA, Charlett A. The rate of diagnosis and demography of
pelvic inflam m atory disease in general practice: England and Wales. In t J
S T D A ID S 1999;10:448-51.
2 Simms I, Stephenson JM. Epidemiology of pelvic inflam m atory disease: what
do we know and what do we need to know? Sex Transm I n f 2000;76:80-7.
3 Simms I, Vickers M , Stephenson J, Rogers P, N icoll A. National assessment of
P ID diagnosis, treatment and management in General Practice: England and
Wales. In t J S T D A ID S 2000;11:440-4.
4 Simms I, Mallinson H , Peeling RW , Thomas K, Gokhale R, Rogers PA, H ay P,
Stephenson JM, Hopwood J. Risk Factors Associated w ith Pelvic
Inflam m atory Disease: A U K Study. Proceedings of the Congress on Sexually
Transmitted Infections, Vienna 2002.
5 Simms I, Eastick K, Mallinson H , Thomas K, Gokhale R, H ay P, Herring A ,
Rogers PA. Associations between Mycoplasma genitalium, Chlamydia
trachomatis and Pelvic Inflam m atory Disease. Proceedings of the Congress on Sexually Transmitted Infections, Vienna 2002.
6 Simms I, Eastick K, M allinson H , Thomas K, Gokhale R, H ay P, Herring A ,
Rogers PA. Associations between Mycoplasma genitalium. Chlamydia
l n l t ’r ih iU i'r : . ! ' /l'ic u iti i>,' , Wf ) s 10 ll-H 4>1
ORIGINAL ARTICLE
The rate of diagnosis and demography of
pelvic inflammatory disease in general
practice: England and Wales
I Simms', P Rogers’ and A Charlett’
^ H IV i‘V STf), C o n u tiu n ic n l'lc S u n v itln n c i' C n t t r r n iu l S tn ii< tic^ U n it, U K
Summary: Kninvled^f of pelvic inllommalory disease (PID) epidemiology is essential to the understanding of roprixiuclive morbidity in women. This paper estimates the rale of PID diagnosis in general practice (C.P) and the level of asscKiation between I'll) diagnosis and demographic factors. Diagnoses of PID were made at 1.7".. of attendances amongst women aged lb to 4b. Increased risk of PID was asscx'iated with smoking ( P < 0.0001), younger age groups (P<0.00t>l) and lower s(K'i(H'conomic groups (P< O.(K)Ol). Compared to patients who were married, increased risk was also associated with those patients who were widowed, separated or divorced and not cohabiting (adjusted rate ratio (RK)=l.b2; confidence limits (Cl.) 1.35 to 1.07), and with those who were unmarried but cohabiting (adjusted KR = I.12; V5".. Cl I II to 1.5b).
General practice is an important ftK'us for the diagnosis and treatment of PID. If intervention and surxeillance are to be undertaken effectively, more has to be known about the epidemiology of this important public health problem
Keyw o rd s; Pelvic in tla m m a lo r) disease, e p id e m io lo g y , rale o f diagnosis
BACKGROUND
Pelvic inflammatory disease, the clinical syndrome associated with upper genital tract infection, is a major health burden in women of reproducti\ e age'. It can cause ectopic pregnancy, tubal factor infertility and chronic abdominal pain, which is asscx'iated with an increased risk of hysterectomy' D ie dominant cause of PID is genital C ltln n iu iH ii tra ch o n u iti^ which is the most common curable sexually transmitted infection (STI) in England and Wales^"*. The recent report by the Chief Medical Officers' Expert Advisory C rou p ' on C. Irn chon uüts
highlighted the urgent need for information concerning the epidemiology of PID for health planning purposes, in particular the assessment of STl intervention programmes. The burden of disease and factors associated with PID are unknown in England and Wales, although the stable, high level of ectopic pregnancy suggests there is a substantial reservoir^. This study estimated the number of PID diagnoses and their asstxiation with demographic and sixioeconomic factors using data from the M o r b id ity S tn tis tii > fro m Genera! P n u 'tice F o u rth N a tio n a l S u n r y : 1991 -1992'
(MSGP4).
C ortv^pentli'tu'c to: Ian Simms, 61 C 'olinJ.ili' .-Vwnnc. I onUon .M\V4 s1;q. L K
METHODS
The MSCiP4 data set" was derived from atten dances over a one-year period 1991/92 at bO CP clinics in England and Walt*s which reprissent a 1% sample of the population. Although not a random simple, it is considered to lx> representative of the general population w ith respect to age, sex, marital status, sociiHVonomic status, smoking behaviour and disease burden'^'. Diagnosi's made by the general practitioner (ICD9 cixles), age, ethnic background, sixioeconomic status, current smok ing habit, the length of time the person was registered at the practice during the year and marital status (a combination variable including marital and cohabitation status) were collected for each patient. One record per person was included in the analysis, records were only included where complete data were recorded. Technically the data are derived from consultations rather than the total age/sex register. However, since 78“o of those on the age/sex register consulted their general practi tioner at least once during 1991/92 this is a close estimate of the burden of disease in the general population, referri'd to here as the diagnostic rate (number of PID d i a gnoses/ person -yea rs at risk). The classification of ethnic group was simplified to: white, black (black Caribbean, black African, black other) and .Asian (Indian, Pakistani, Bangladeshi). The analysis included 73,810 women aged 16 to 46.
S m in iv . Kii>;ctv .itu l I h .irlc tl I ’ r o . i lt - i u i - ili-n iu ^ r.ip h v t>( I ’ l l ) 11*^
hil'lt- I f'l'/rii inll,iiiinhili'Hf if>( i lli.tMHI j f n-~k). riUc riitii*'. U iK -) ,iiui '• ‘i"., i iiufuirihi- In tii/- <(. I >
IV is .11, lu f (i . w s ll) iKkl -\<l(iisitH.I KK / ’ l . l l t l f \ . ir i.il'lf [STsiin i i'.irs .11 risk i e ' V ' ,1 1 1 (.u l|u s t,\l K K l
\>:«- .virmij'
Is to l ‘i 22V ("1 liC 2 l 1 DO ■ ODOOl : n to 21 2 Ï1 (2.SU I1.V21)) 1 1)7 III .S4 1 Vs)
2 i to :>» 221) (5 n s 1 V.u— ) ( I'M 11)72 1 21) VI) to VI is s ,2M |V ,2 2 1) (1 so ( i ) s i 1 ()V) VV to VI 1 2 : 11 W 1 l.iv v ,) (1 V4 (1 )4 0-41 7 2) ID to -It, s v i i n v ii,.v v 4 ) 0 2s (1) !>' 0 4s) 1 tlin iiilv W l u l f w r m t v i s 't.s 't:. 1 IM) 0 0 W 4 I5l.uk 2 » s jii i V2'ii I sV (0>»7 2 TV)
A ii.iii 1 S I 111 Si.Vj 1 VV (0 ,s.| 2 7S) M .in t.il s l.ih is
\1 .irru s l |V | |V |,I 41.1251 1 IH) ■ 0 (4 )0 1 V o h .ilu liiij; 2 i - 2 i n t , :4 t,2 i 1 5 2 ( 1 1 1 1 4SI
IV u io m s t M - p ,ir.ilfi( 2 It, 11 5.S V,S40| 1 s2 (1 VV I „ 7 i i ( i i o r .i N i n ot io li.il> ilin » : I ‘ |2 1 5 |s In, (t.Vl OSS 1(1 74 1 0 4 ) S.X i.il 1 l . e - I II 114 (22 5 l'»,V'v4| 1 IW) - OlKIOI I II n o n n i.in u .ll | V - ( | | V 2S2«X)| 1 22 I I 04 1 4 5) I II n i.in u .ll 225 (1 VI V s,,:) 1 4s I I 2s 1 '4’ ) IV p .ir iu s k ill, a 2Vv ( V I 1 1 V 1211 1 s 4 11 ys 1 ,j7) V i" s ( : s v ts 't| 1 4 2 1 1 17 1 VS) s in o kin p . ) , s 2*y, (VUH 22 4WD 1 s4 1 1 s 4 2 IM ) \ o 122 (V ')l 4s 2 ^ 0 1 DO ' • ( n ' o u ï
H u ‘ liiin ' p.itu'iits w i-rr n-^isii'n'vl u ill) (ht ir p M it'r.il p r .iit ilitin iT \ . i r m i Irm n ,i tla \ (n a \i*a r
I h f lU 'n o m in a ti'r was thus la lc u la ti'tl as p u rsttn - w a r s a I hsk In ih v id in i; tho m iu ih rr of i la \s uat h person w as iiu liK ied in (lit' s ii r \e \ h \ ItVi ( I w as a leap \4>ar) ^ i\m y ; a lo lai ot 711.7*^)I p e rs o n - years at risk Siny;le am i n u iitn ariah le analyses w ere u n d e rta kei) usin^ a I’t'isson rey;ression in m lel (S l A I A 5 0). I he outeom e \ a ria h le w as a diaypiosis ot r i i ) m ad e iliiriny; the s tm h perio d (R l> ) t ode (>14). N o distin ctio n was m ad e betw een w h eth er these w e re new diai^noses o r w ere a consultation t«>r an ep iso tle lirst tiiagnost'd »>utsiile (he s tiu lv perio tl I he analysis thus y;i\ es a prev alence estim ate Interactions w » re in vestigated and a m ain ettects m odel w as used to ilest rihe the data U nad ju sted and ad jiistetl rate ratios (KKs) w ere calculateil
in SUI TS
The rate ot I’ l l ) diay;noses in w o m e n ay;etl l(> to 4(> altendiny; CTI’ was lt>7 per 1 0 / X X I pe rs o n -vea rs at risk ( 1 IHM/70,7^M ), or 1.7",.. The n u m b e r t't diay;nos»>s and the rate pe r 1 0 , ( X X I person ve.irs at risk and TvKs adjusted tor other variables in the rey;ression analvsis to gethi'r w ith C I are sh ow n in T able I. I he via ta w ere re-coded to av oid prob lem s w ith sparse data in some calegivrii-s. .All 2 wav interactions w \ re mv estiy;atei.l but none w ere
to un ti to be siy;niticant at the lev el There was a sig nificant ditteren ce h e tw iv n age grou ps, w ith w o m e n aged 4s to at halt the risk ot h a vin g a d ia g n o s is o t I ' l l ) (O 'O tXX)!, adjusted K K=0.44, MV'., ( I 0.40 to 0,72) and those aged 40 to 4(> w ere at a vjuarter ot the risk (adjusted KR-0.2(>, 11 O b) to 0,4(1) co m p ared to the l(> to b(-vear age gri>up S m okers w ere at higher risk of I ’ l l ) than non- sm okers ( /'< 0 .(X X ) |, adjusted RK - US, Os"„ CT l.(>5 to 2 tW). I'atients in sot u n v o n o m ic g rou ps III to \ w ere all at h ig h e r risk ot I’ l l ) than those in s(x iiH 'io n o m ic g ro u p I II ( / ’ < 0.(XH)l ). ComparetT to patien ts w h o w e re m a rrie d , increased risk of I’ l l ) w as also assiK lateif w ith those patien ts w h o w ere w id o w e d , separated o r d iv o rc e tl and not co hab it ing (atljusted R R = l.(i2 , Vs".. C l I 4s to 1.V7), and w ith those w h o w ere u n m a rrie d but co hab itin g (adju sted R K = I 42, V s " ,. C'l I I I to I ,s(>). The d ifféren ce in the risk of I ’ l l ) b e tw ee n ethnic groups w as not statistically sig nificant f / ’= 0 .W 4 ), but there w as ev idence ot increased risk in l>oth black (adjusted R R = l.(o , V s " „ y | lt.v7 to 2 7V) and Asian patien ts (adjusted RR=I7>4. V s " „ C l (f.S4 to 2.78) co m p aretl to Caucasians
D IS C U S S IO N
I he e p id e m io lo g y of I ’ l l ) is n o to rio u s ly d iffic u lt to study a n d , to ou r k n o w le d g e , this is the firsf inv estigalion to be u m le rla k e n in T n g la n d and
4SI) In ii‘m «itioiw l )t'iirn.»l ni S i l ) (St AIDS V u liim n III |iily
Wtiles. This is, in part, due to the heterogeneity in the pattern of disease, the low specificity of clinical diagnosis, and the fact that women with PID present to a variety of different clinical spsx'ialities. in addition, it is becoming increasingly clear that many cases of PID go unrecognized because they are atypical or asymptomatic The epidemiology of PID is also thought to vary with microbial aetiology which changes over time. There are thus a number of difficulties in both interpreting the surveillance data presented here and in comparing the results with other studies. However, the lack of published PID surveillance data for Hngland and Walt's indicates that this was a valid investigation to undertake and can be used as the basis for future validation studies.
The 1.7".* rate of PID diagnosis suggests that l63,tXX) cases of PID would have been diagnosed in C P among reproductive age women in 1492. This contrasts with only 21,168 and 57T5 cases seen as hospital inpatients'” and altenders at genitourinary medicine clinics" respectively in the same year. Although the case definition of PID is likely to have va ril'd between settings, these data suggest that a sul>stantial reservoir of PID is seen in general practice. This clinical setting should thus be an important focus for the diagnosis and treatment of PID.
O ur study supports previous obsen ations that women who smoke are at significantly higher risk of P ID ’’ . Although hospital inpatient data indicate that women aged 20 to 24 years are at highest risk of PID' ' this multivariable analysis indicates that women betwivn 16 and .T4 years are at i\;ual risk. This shows that, although highi*st rates of genital chlamydial infection pt'aks in teenage women', the morbidity associated with chlamvdial inflation affects the reproductive health of women o\ er a substantial age range.
Recent studies have speculated whether black ethnic groups have a high burden of repnxluctive m orbidity'''"’, black and Asian ethnic groups are under represented in this analysis, accounting for only 1.6"i. of the study population, compared to 5",. reported in the 1991 census’ After adjusting for sociix'conomic status, these groups were not found to be at higher risk of PID than whites. However, the higher RRs and confidence intervals only just encompassing one indicate that risk of PID was approaching significance for blacks and, to a lesser extent, Asians. Clearly the burden of PID in different ethnic groups needs to be explored using larger studies.
Data on sexual behaviour are central to the study of a predominantly sexually transmitted disease such as PID. A relationship between risk of PID and divorced marital status has also been reported in studies in England and Wales'^ 'T This associa tion and that between PID and socioeionomic group are probably surrogate markers of sexual behaviour. I or example, age at sexual debut and number of lifetime sexual partners are known to
vary with marital status, cohabiting and socio^ economic g ro u p U n fo r tu n a te ly sexual behaviour data were not included in the MSGP4 data set; further studies using measures of sexual beha viour, are required to validate these findings.
Knowledge of the epidemiology of PID is central to the reprixluctive health of women. We have shown that P ID affects women over a broad age range, particularly t h i ^ who smoke, are in lower sixrioeconomic groups, or are divorced or sepa- rati'd. There is also some evidence of increased risk in blacks and Asians. I lowever, a number of problems have to be resolved if an accurate view of the epidemiology of PID is to be made. A standard case definition should be developed for future research, the diagnostic rate given here should be validateil and the factors assiKiated with PID need to be investigated. These are difficult problems to resolve, but it is only by doing this that epidemiological knowledge of this important public health problem w ill be improved.
C o n trih iitiiy n o f lu itlio m : I Simms instigated the project, retrieved and edited the data and lixik the lead in w riting the paper. P Rogers and A Charletl advised on study design, statistical analvsis and wrote the statistics section.
A iA'/niîr/cdyr//f<7(ts.- We would like to thank D r C Bevan (Consultant, Weslon-Super-Mare General Hospital), Dr A Swan and M r A Grant (PHLS Statistics Unit) for statistical advice and technical assistance.
References
1 W u rlit ll.Mik. IVi'rM K.'forf /wc.'s/ric in H riillh . t)x (o rd : t K l o r t I U n i\c r> ily 1W3
2 D iiih .in I t . V'i*sM-v M , t io lila ir c M , I jir w iM lh iT ). M orl>iitity foll.'vvinj; jx'K ic infl.>mni,it«*rv (list .ist- rtv o n l linluigc sliitlv. /t* I |441;100:sSS 62
.) C D , Ml, ( i, K id ^ w .iy C l., Siddlc N C , Clinit'.il, iapjrnscopic and niu'rnt’ iologii'.i! (indings in acute s.ilpingilis: report on a U n ited K in g d o m cohort. Mr /
GuminolMWx;102:407- I4
4 Sim m s I, C atch pole M , B n ig h a R, Rogers P, Vtallinson I t . \ i c o i i r.p id em io lo g v of genital Chlnrnyilid trdclhnnatisin I ngi.ind and W ales. CniiUniritiiVfol lV9 7;73: l22 -6
5 D eparlrnent o f H e alth . C h h m y ilM Sum nM ry iitu i
Con*/lO/iiiis y f C A t()'s r .x iv r l A J i'in n v Crou(>. lo n d o n : IX 'p a rlm e n t o f H e a lth , PWS
h Sim m s I. Rogers P A , N ic o ll A . l l i e influence o f dem»v graph ic change and cu m u la tiv e risk of pelvic inflam m atory
d is e a s e on llie incidence of ectopic pregnancy, t'/n tft'n iio l
l»l,-G | W7;119:4‘» -S2
7 CHfice of Population Censuses and Surveys. M o rM ity