Management of first-episode pelvic inflammatory disease in primary care: results from a large UK primary care database

Abstract

Background

Prompt and effective treatment of pelvic inflammatory disease (PID) may help prevent long-term complications. Many PID cases are seen in primary care but it is not known how well management follows recommended guidelines.

Aim

To estimate the incidence of first-episode PID cases seen in UK general practice, describe their management, and assess its adequacy in relation to existing guidelines.

Design of study

Cohort study.

Setting

UK general practices contributing to the General Practice Research Database (GPRD).

Method

Women aged 15 to 40 years, consulting with a first episode of PID occurring between 30 June 2003 and 30 June 2008 were identified, based on the presence of a diagnostic code. The records within 28 days either side of the diagnosis date were analysed to describe management.

Results

A total of 3797 women with a first-ever coded diagnosis of PID were identified. Incidence fell during the study period from 19.3 to 8.9/10 000 person-years. Thirty-four per cent of cases had evidence of care elsewhere, while 2064 (56%) appeared to have been managed wholly within the practice. Of these 2064 women, 34% received recommended treatment including metronidazole, and 54% had had a Chlamydia trachomatis test, but only 16% received both. Management was more likely to follow guidelines in women in their 20s, and later in the study period.

Conclusion

These analyses suggest that the management of PID in UK primary care, although improving, does not follow recommended guidelines for the majority of women. Further research is needed to understand the delivery of care in general practice and the coding of such complex syndromic conditions.

INTRODUCTION

Pelvic inflammatory disease (PID) is a clinical syndrome involving abdominal pain and tenderness, dyspareunia, abnormal bleeding, discharge, and fever. It is predominantly the result of a bacterial (often sexually transmitted) infection ascending from the endocervix to the higher reproductive tract. PID can lead to long-term complications including tubal infertility, ectopic pregnancy, and pelvic pain and it is thought that prompt treatment reduces the risk of such sequelae.¹ UK guidelines for the treatment and management of PID cover appropriate testing and antibiotic treatment and include suggested outpatient regimens.^2–5 Although some women in the UK experiencing the symptoms of PID may seek help in specialist clinics, such as genitourinary medicine clinics, or go direct to secondary care via emergency departments, many are likely to attend their GP.^6,7 Indeed, analysis of records for patients consulting in general practice have indicated that a substantial and increasing proportion of sexually transmitted infections (STIs) are dealt with in primary care settings.^8,9 This trend follows England's National Strategy for Sexual Health and HIV, which recommended a greater role for primary care providers in the detection and treatment of STIs.¹⁰

Little is known about how women with PID are managed in general practice, or how closely management guidelines are followed in this setting. The existing literature is based on self-report questionnaires, which often have very low response rates and are subject to reporting bias.¹¹ However they show that guidelines are often not followed. Chlamydia trachomatis is an important causative agent of PID and concern about the burden of C. trachomatis infection in men and women aged 15–24 years and the potential long-term sequelae resulted in the National Chlamydia Screening Programme (NCSP) in England.¹² This programme offers opportunistic screening from various clinical and non-clinical sites. It is not clear how this programme has affected the management of PID. There is a need for recent data to describe delivery of care and inform continuing education.

Research databases of electronic primary care records enable identification of women seen in general practice and given a diagnostic code for PID. The presence of such a code may not always accurately reflect the presence of the clinical condition, but the large body of varied research using these resources suggests that they allow valid examination of real-life patient journeys.

Research objectives

Using electronic primary care records, the study aimed to estimate the incidence of first-episode PID seen in general practice by identifying a cohort of women consulting with a first-ever episode of PID. It also aimed to describe the management of these patients, within the practice and beyond, and assess its adequacy in relation to existing guidelines, including associations between management and various patient and practice factors.

METHOD

Data used

The General Practice Research Database (GPRD) is an electronic database of anonymised longitudinal patient records from general practice.¹³ Established in 1987, it is a UK-wide dataset covering 5.5% of the population, with data from 460 practices, broadly representative of the UK population. There are 3.5 million currently active patients. Records are derived from the GP computer system (VISION) and contain complete prescribing and coded diagnostic and clinical information, as well as information on tests requested, laboratory results, and referrals made at or following on from each consultation. The structure of the data is shown in Figure 1, with different parts of data held in separate record tables. Practice-level data include a deprivation index score based on the postcode of the practice and the NHS region in which the practice is based. In 200 out of 460 (approximately 40%) practices, a Townsend deprivation index score based on the small area level is available for each patient.

Figure 1.

Structure of the GPRD database.

How this fits in

General practice has an important role in treating pelvic inflammatory disease. Management guidelines exist, but it is uncertain whether these are followed in general practice. Coded data from an electronic database of real-time primary care records indicate that a minority of women with pelvic inflammatory disease receive investigation and treatment according to guidelines. Further research is needed to understand the delivery of care in general practice.

Identification of cases

The target population was all women permanently registered with a practice that met GPRD data reporting standards. All women with a first coded diagnosis of PID, who were aged 15–40 years at the time of diagnosis, during the study period 30 June 2003 to 30 June 2008, were identified. Code lists used for the definition of cases are included in Appendix 1. The code lists focused on acute cases, so that women with a code for chronic PID or its sequelae before the first acute code were excluded. Because the study research question concerns management, only codes that the researchers thought indicated that the GP was confident of the diagnosis of PID were used. Codes where the diagnosis may have been in doubt such as ‘female pelvic infection’ were excluded. Women with a coded diagnosis relating to pregnancy, miscarriage, or therapeutic abortion 60 days before to 28 days after the date of the PID code were not included in the analyses, and nor were temporary patients.

If there were multiple PID diagnostic codes for an individual, the date of the first one was used as the index date. Analyses were restricted to records in the period 28 days before and after the index date. If the index date was within 28 days of the start or end of the registration at the practice, the case was excluded from the final management analyses.

Analysis of management

Four aspects of the management of the PID patients were assessed.

Testing

A chlamydia test was considered to have been carried out if there was either a code for a test (for example, ‘chlamydia antigen test’) or a diagnosis of relevant chlamydial infection (for example, ‘female chlamydial pelvic inflammatory disease’), using an approach developed in previous work.⁹ Codes were identified for tests for Neisseria gonorrhoea. Non-specific microbial tests were considered to have been carried out if there was either a code for appropriate swab (for example, ‘high vaginal swab’), or a test such as microscopy, culture, and sensitivities with no location given. Code lists are included in Appendix 1.

Treatment

Using prescription data, variables were created for each of the following treatments:

• all antibiotics – based on British National Formulary heading 0501; and

• specific antibiotics – based on guidelines (2005 and 2006),^3–5 recommended treatment for PID was defined as any one of erythromycin, ofloxacin, azithromycin, or doxycycline, with or without metronidazole. Code lists were drawn up using drug substance name and including all formulations except for inappropriate topical preparations. Dosage and duration of use were not assessed, since this is a complex task and resources were limited. Although it is optional for milder cases in guidelines, the use of metronidazole has been highlighted in these analyses as a marker of the most complete treatment.

Evidence of care elsewhere

It was considered that a woman with a diagnostic code for PID had received care for PID in another healthcare setting if any one of the following conditions were met:

• a diagnostic code for the condition within the referral record (Figure 1);

• a suggestive symptom code within the referral record (for example, ‘pelvic pain’); or

• a code anywhere in the records indicating care elsewhere (for example, ‘referral to A&E’, ‘seen in GUM clinic’). This category also included less specific terms such as ‘discharge summary’ or ‘letter from specialist’.

Evidence of management only within practice

If there was no evidence of care elsewhere and there was evidence of any treatment (any antibiotic prescription) or testing (including non-specific microbial tests) within the practice, these cases were considered to have been managed within the practice only. This group alone was used for the investigation of quality of management in general practice as it did not seem appropriate to include cases where important parts of the care may have been delivered outside the general practice, and hence not necessarily recorded on the database. Women with no evidence of either management within the practice or care elsewhere were also not included in the quality of management analyses, due to concerns about completeness of recording in these cases.

Statistical analysis

The data were prepared using Stata (version 10; Statacorp LP, Texas). Calendar years were defined as mid-years from 30 June so that year 2003 covered 30 June 2003 to 29 June 2004, and so on.

Denominator data, by single year age groups and calendar year, were used to calculate the incidence rates in specific age groups and years by dividing the number of cases by the appropriate denominator. Age-standardised rates were then obtained by applying these rates to the European standard population. Differences in incidence rates over time and age groups were assessed using Poisson regression.

For analyses of management, first the proportion of cases with the various management markers in different study years and age groups was calculated. Logistic regression models were then used to investigate factors associated with different management locations, testing, and treatment patterns. A series of sensitivity analyses were performed, extending the window for analysis of management from 28 to 42, 60, and 90 days either side of the index date, to assess whether relevant data were being missed by using the 28-day window.

RESULTS

Figure 2 summarises the identification and exclusion of cases; 4724 women with a first coded episode of PID during the study period were identified from a population of 2.8 million woman-years of observation. Of these, 927 had evidence of a recent pregnancy and were excluded, leaving 3797 cases for the incidence analyses. The median age of the included women was 25 years (interquartile range 20–31 years).

Figure 2.

Flow chart of study: patient identification and exclusions.

Incident cases available for the study

The incidence of first episode of PID seen in general practice fell during the study period, from 19.3/10 000 person-years in 2003 to 8.9/10 000 person-years in 2007. The incidence rate was highest in women aged 20–24 years, and rose from 28.4 to 30.4/10 000 person-years between 2003 and 2004 but then fell to 14.2/10 000 person-years in 2007. Incidence in all other age-groups fell progressively from 2003 to 2007, with the greatest decline in younger age groups (P = 0.04 for interaction in Poisson models).

Management of PID

Location of management One hundred and twenty-eight women in whom the index date of the first episode was within 28 days of the start or end of their GP records were excluded, leaving 3669 cases of PID (Table 1); 34% of women with a PID diagnosis had evidence of having received care elsewhere, and this proportion increased during the study period (P for trend <0.001). To validate the classification of location of management, the proportions of women who received treatment or testing within the practice were compared in women with and without evidence of care elsewhere. Evidence of care outside general practice was associated with a 40% reduction in odds of having either recommended treatment or a C. trachomatis test within general practice (odds ratio [OR] = 0.6; 95% confidence interval [CI] = 0.5 to 0.7 for both). The proportion of cases with evidence of management only within the GP practice declined over time (P<0.001), commensurate with the rise in cases with care elsewhere. Ten per cent of women had no evidence of either management within practice or care elsewhere, and this proportion also declined over time.

Table 1.

Location of management of PID cases seen in primary care.

		n (%)
	n	Evidence of care elsewhere	Managed only in practice	No evidence of management
2003	954	247 (25.9)	579 (60.7)	128(13.4)
2004	892	283 (31.7)	528 (59.2)	81 (9.1)
2005	754	267 (35.4)	415(53.0)	72 (9.6)
2006	605	235 (38.8)	317 (52.4)	53 (8.8)
2007	464	202 (43.5)	225 (48.5)	37 (8.0)
P-value for trend over time		<0.001	<0.001	0.001
Total	3669	1234 (33.6)	2064 (56.3)	371 (10.1)

Management within practice in comparison to guidelines

Of the 2064 PID cases managed only within practice, 60% of these women received recommended treatment (of which 34% included metronidazole) (Table 2); 56% had received a C. trachomatis test, and 50% a non-specific microbial test; 36% received optimal treatment with the recommended antibiotic and a C. trachomatis test, but only 16% had had a C. trachomatis test and treatment including metronidazole. The absolute numbers of cases managed within practice declined steeply during the study period, due both to declining overall incidence and the increasing proportion with evidence of care elsewhere. The proportion of women receiving recommended treatment increased during the study period, but there were no clear trends in testing over time. Sensitivity analyses, shown in Appendix 2, show that the proportion of cases with evidence of care elsewhere increased as the window widened, but for patients managed within the practice, the pattern of care was similar.

Table 2.

Quality of management of PID cases managed within practice only.

		n (%)
	N	Chlamydia trachomatis test	Non-specific test	Neisseria gonorrhoea test	Recommended treatment	Recommended treatment (including metronidazole)	Recommended treatment + Chlamydia trachomatis test	Recommended treatment including metronidazole + Chlamydia trachomatis test
2003	579	306 (52.9)	299 (51.6)	4 (0.7)	331 (57.2)	195(33.7)	180(31.1)	74(12.8)
2004	528	297 (56.3)	246 (47.0)	3 (0.6)	304 (57.6)	161 (30.5)	180(34.1)	67(12.7)
2005	415	243 (58.6)	210(50.6)	14(3.4)	241 (58.1)	128(30.8)	150(36.1)	64(15.4)
2006	317	181 (57.1)	157(49.5)	9 (2.8)	210(66.3)	126(39.8)	130(41.0)	64 (20.2)
2007	225	130(57.8)	114(50.7)	20 (8.9)	152(67.6)	98 (43.8)	97 (43.1)	51 (22.7)
P-value for trend over time		0.116	0.914	<0.001	0.001	0.003	<0.001	<0.001
Total 2064(100)	1157(56.1)	1028(49.8)	50 (2.4)	1238(60.0)	708 (34.3)	737 (35.7)	320(15.5)

Patient and practice-level associations with optimal treatment

Determinants of optimal treatment (C. trachomatis test with recommended treatment with or without metronidazole) were examined in logistic regression models. Multivariate models confirmed optimal treatment was more likely in later years of the study. Women in their 30s were less likely to receive optimal treatment (Table 3). Although there were some differences in optimal management across the quintiles of practice deprivation, there was no evidence of linear trend with increasing deprivation (OR for linear variable = 1.0 [95% CI = 0.9 to 1.0]). Similarly, no association with deprivation was seen in the subsample for which individual Townsend deprivation index quintile was available (n = 982, P from likelihood ratio test for inclusion of deprivation quintile = 0.55). Results were similar when multilevel logistic regression models were run with a random intercept including practice as a higher-level variable.

Table 3.

Determinants of optimal management^a within practice.

	Age group (years)
	15–19	20–24	25–29	30–34	35–40
Adjusted OR (95% CI)	1	0.9 (0.7 to 1.1)	0.6 (0.5 to 0.8)	0.5 (0.3 to 0.7)	0.4 (0.3 to 0.6)
Event year	2003	2004	2005	2006	2007
Adjusted OR (95% CI)	1	1.1 (0.9 to 1.5)	1.2 (0.9 to 1.6)	1.6 (1.2 to 2.2)	1.8 (1.3 to 2.5)
Practice location: quintile of deprivation	1 (least deprived)	2	3	4	5 (most deprived)
Adjusted OR (95% CI)	1	1.5(1.1 to 2.1)	1.3 (0.9 to 1.8)	1.6 (1.2 to 2.2)	1.1 (0.8 to 1.5)

^aOptimal management = C. trachomatis test and recommended antibiotic treatment with or without metronidazole.

DISCUSSION

Summary of main findings

The study results confirm that a substantial caseload of PID is seen and diagnosed in UK primary care. A fall in incidence of diagnosed cases was seen between 2003 and 2008. The results indicate, however, that only a minority of PID cases managed within primary care received treatment and investigations according to guidelines. Overall, only 16% of women received both a C. trachomatis test and recommended treatment, including metronidazole. Women seen later in the study period were more likely to have received care that followed guidelines (23% received testing and treatment including metronidazole in 2007), as were younger women. Practice-level deprivation did not affect the recorded management of patients.

Strengths and limitations of the study

This study used real-time patient records and so it was possible to assess the care that unselected patients receive, avoiding the problems of response and reporting bias that are inherent in self-report data. However, these databases are designed for patient care and not for research. Only coded data were used in these analyses, and it was not possible to include any information that was entered as free text in the GP records. This may have resulted in misclassification of patients both as cases and into the different management groups.

The clinical diagnosis of PID is subjective, with definitive diagnosis relying on laparoscopy which is not routinely used. Doxanakis et al showed substantial differences in the diagnostic rates of PID between genitourinary medicine specialists in Australia.¹⁴ This is also likely to be true between GPs. Once a diagnosis is made, GPs may vary in their use of Read Codes, so that a condition may be entered either as diagnosis or as presenting symptom. In order to maintain high specificity, no presenting symptoms were included as diagnostic codes. Ratelle et al showed that PID diagnostic codes alone in secondary care were a poor predictor of clinically verified PID.¹⁵ If this is also the case in primary care, some of the cases in the present study may not have actually been PID. Although there are no validation data to address this issue, the key question for this study is whether the GP thought the patient had PID and whether they were then managed appropriately; such as, whether the diagnostic code accurately reflects the GP's opinion. Further research is needed to understand how GPs code such complex clinical conditions.

The classification of location of management was complex. The use of the referral record by GPs was used as evidence of referral, but it is unclear how accurately these separate records are used in practice by GPs. Some of the Read Codes taken as evidence of care elsewhere were relatively nonspecific and may have not actually been related to the PID diagnosis. As expected, as the management window widened, the proportion of women with evidence of care elsewhere increased, but this may have included unrelated referrals. Similarly, it is possible the recommended treatment was given for conditions other than PID. However, in sensitivity analyses the estimates of treatment were not dependent on the length of the management window.

Comparison with existing literature

Previous papers reporting incidence of PID in UK primary care have studied total episodes not just first episodes. Simms and coworkers, using two different primary care systems, reported incidence rates up to 10 times higher than those of the present study: with 251/10 000 person-years for 20–24 year olds in 1991, and 150/10 000 for 25–44 year olds in 2001.^6,7 Other work using the GPRD has found an incidence of 28.1/10 000 person-years for women aged 16–44 years between 2000 and 2008.¹⁶ The incidence levels for PID reported in this study are lower because the incidence estimates are not for surveillance but to estimate management case load in general practice. A more restricted code list was used to increase specificity, first episodes only were counted, and cases where the infection may have arisen in relation to childbirth or miscarriage were excluded.

The study found a marked fall in the incidence of first-episode coded PID in primary care, with rates falling by more than 50%. Declining incidence of PID diagnosed in general practice in the UK has been documented since 1994,7 and its relationship to the English NCSP is at yet unclear.¹⁶ The study data may reflect a true fall in incidence of the disease, but alternative explanations need to be considered. These include changes in patterns of care – with more cases seen in other settings – or changes to coding or recording of data. Recent changes in GP coding and recording behaviour have been substantial, following the introduction of the Quality and Outcomes Framework targets. It is possible that such changes have led to an apparent fall in PID incidence. However, PID is not included in any Quality and Outcomes Framework targets, and for improved coding to lead to a fall in incidence would imply that previous estimates were too high. The trend found of increasing evidence of care elsewhere might support change in the pattern of care as an explanation. However, reports from other countries that include data from all healthcare settings have shown a similar trend with falling incidence.^17,18 If incidence is falling, it may be due to the chlamydia screening programme or may reflect a change in pathogenicity of causative organisms.

Implications for clinical practice and future research

The results overall raise some concerns about the completeness of management of PID in primary care, although the trend of improved management over time is encouraging. Although 70% of women with PID had received recommended treatment and 54% a C. trachomatis test, only 16% had received both. A postal questionnaire survey of UK GPs in 1994 found that 63% reported prescribing appropriate anti-chlamydia treatment and 45% reported testing for C. trachomatis.¹¹ Overall, these figures are very similar to the estimates of the present study, and suggest education could still be improved. Women in their 20s are more likely to receive recommended management. It was reassuring that neither patient nor practice deprivation level were associated with following management guidelines.

Cases of PID recorded in primary care were more likely to show evidence of care received elsewhere in the later years of the study period. This trend towards greater care elsewhere is somewhat surprising and goes against what might have been expected from the recommendations of the Department of Health review.¹⁰ It is possible that these trends represent better recording of referral and hospital events by GPs rather than any actual change in pattern of care.

These analyses, using coded data from an electronic records database, suggest that the management of PID in UK primary care, although improving, does not follow recommended guidelines for the majority of women. Further initiatives are required to improve the management of women with PID and to understand any barriers to care. However, the accuracy of the coded information in primary care databases needs to be confirmed and the authors plan to consult anonymised free text in a selection of cases to investigate whether textual data alter the case definition and estimates of incidence and management.

Appendix 1. Code lists: A. Diagnostic codes for pelvic inflammatory disease

GPRD medical code	Read/OXMIS term	Read/OXMIS Code
271330	Female pelvic inflammatory disease NOS	K4z..00
216451	Other specified female pelvic inflammatory disease	K4y..00
262041	Female gonococcal pelvic inflammatory disease	K44..00
225561	PID – pelvic inflammatory disease	K40z.13
243685	PID	K40z.11
271318	Female pelvic inflammatory diseases NOS	K40z.00
207468	Female chlamydial pelvic inflammatory disease	K40y100
289486	Acute pelvic inflammatory disease	K409.00
262035	Acute pelvic inflammatory disease	K406.11
262034	Acute parametritis	K403000
234667	Acute parametritis and pelvic cellulitis	K403.00
252797	Unspecified salpingitis and oophoritis NOS	K402z00
225555	Perisalpingitis unspecified	K402700
207465	Salpingitis unspecified	K402600
252796	Salpingo-oophoritis unspecified	K402500
207464	Perioophoritis unspecified	K402400
289481	Oophoritis unspecified	K402300
280371	Salpingitis and oophoritis unspecified	K402.00
262032	Acute salpingitis and oophoritis NOS	K400z00
216441	Acute perisalpingitis	K400400
234664	Acute salpingitis	K400300
298752	Acute salpingo-oophoritis	K400200
271313	Acute perioophoritis	K400100
207462	Acute oophoritis	K400000
298751	Oophoritis – acute	K400.11
280367	Acute salpingitis and oophoritis	K400.00
234663	Ovarian, fallopian tube and pelvic inflammatory diseases	K40..00
243681	Female pelvic inflammatory diseases	K4...00
269816	Acute gonococcal salpingitis	A981700
304410	PELVIC INFLAMMATION FEMALE	6160FL
238388	PERIOOPHORITIS	614 PP
238387	PERISALPINGITIS	614 PL
284301	OOPHORITIS	614 B
304403	SALPINGO-OOPHORITIS	614 A
211327	OOPHORITIS ACUTE	612 P
229476	SALPINGITIS ACUTE	612 L
219863	OOPHORITIS GONOCOCCAL	0981B
256278	SALPINGITIS GONOCOCCAL	0981A
207522	[X]Inflammatory diseases of female pelvic organs	Kyu8.00
214967	Chlamydial inf of pelviperitoneum oth genitourinary organs	A78A300
265498	SYNDROME FITZHUGH CURTIS	0988FC
278875	Fitzhugh Curtis syndrome	A98y600
289351	Chlamydial peritonitis	J550400

Appendix 1.

Code lists: B. Code lists for chlamydia test.

GPRD medical code	Read/OXMIS term	Read/OXMIS Code
205965	Chlamydial infection, unspecified	A78AW00
205969	Other viral or chlamydial disease NOS	A7z..00
206063	[X]Other chlamydial diseases	Ayu6100
207468	Female chlamydial pelvic inflammatory disease	K40y100
214967	Chlamydial inf of pelviperitoneum oth genitourinary organs	A78A300
215059	[X]Chlamydial infection, unspecified	Ayu6200
225563	Chlamydia cervicitis	K420900
242170	Chlamydial infection of genitourinary tract, unspecified	A78AX00
242258	[X]Chlamydial infection of genitourinary tract, unspecified	Ayu4K00
251351	Chlamydial infection of lower genitourinary tract	A78A000
258276	Chlamydia antigen by ELISA	43U0.00
267536	Chlamydia antigen test	43U..00
278838	Other viral and chlamydial diseases	A7...00
278847	Other viral or chlamydial diseases	A78..00
278852	Chlamydial infection	A78A.00
280340	Chlamydial epididymitis	K241600
285745	Chlamydia antigen ELISA positive	43U1.00
285746	Chlamydia antigen ELISA negative	43U2.00
287974	Other specified viral and chlamydial diseases	A78y.00
289351	Chlamydial peritonitis	J550400
297184	Chlamydial infection of anus and rectum	A78A200
297190	Other specified viral or chlamydial diseases	A7y..00
297288	[X]Other diseases caused by chlamydiae	Ayu6.00
302966	INFECTION CHLAMYDIAL	0399C
302967	CHLAMYDIA TRACHOMATIS	0399CT
307938	Chlamydia trachomatis IgG level	43eJ.00
308079	Chlamydia trachomatis L2 antibody level	43eC.00
308199	Chlamydia group complement fixation test	43eF.00
308461	Chlamydia antibody level	43eE.00
308950	Chlamydia trachomatis polymerase chain reaction	43h0.00
309472	Chlamydia group antibody level	43WM.00
309613	Chlamydia trachomatis IgM level	43ez.00
309766	Endocervical chlamydia swab	4JK9.00
309829	Urethral chlamydia swab	4JKA.00
332003	Chlamydia trachomatis IgA level	43n9.00
342066	Chlamydia trachomatis antigen test	43U3.00
342214	Chlamydia deoxyribonucleic acid detection	43jK.00
342310	Chlamydia serology	4JDM.00
343726	Urine screen for chlamydia	68K7.00
343949	Chlamydia PCR positive	43U4.00
343968	Chlamydia PCR negative	43U5.00
344624	Urine chlamydia trachomatis test positive	46H6.00
344736	Urine chlamydia trachomatis test negative	46H7.00
345942	Chlamydia screening declined	8I3T.00
346998	Chlamydia screening counselling	677L.00
347186	Chlamydia trachomatis contact	65PJ.00
347227	Low vaginal swab for chlamydia taken by patient	4JKD.00
347301	Chlamydial infection of genital organs NEC	A78A500
347315	Chlamydia test offered	9Oq0.00
347970	Chlamydia test positive	43U8.00
348085	Chlamydia test negative	43U6.00
348329	Chlamydia test equivocal	43U7.00

Appendix 1.

Code lists: C. Tests for Neisseria gonorrhoea.

GPRD medical code	Read/OXMIS term	Read/OXMIS Code
249090	Gonorrhoea infect. titre test	43E6.00
309228	Neisseria gonorrhoeae polymerase chain reaction	43h6.00
309635	Neisseria gonorrhoeae nucleic acid detection	43jA.00
340376	Gonococcal swab	4JLA.00
342356	Gonococcal cervical swab	4JKB.00
343558	Gonococcal urethral swab	4JKC.00
348093	Gonorrhoea test positive	4JQA.00
348168	Gonorrhoea test negative	4JQ8.00
348381	Gonorrhoea screening counselling	677M.00

Appendix 1.

Code lists: D. other microbial tests.

GPRD medical code	Read/OXMIS term	Read/OXMIS Code
203712	Infectious titres NOS	43E..00
203917	Sample microscopy	4I15.00
203918	White cells seen on microscopy	4I15100
203919	RBCs seen on microscopy	4I15200
203947	High vaginal swab culture negative	4JK2100
203948	HVS culture – trichomonas vaginalis	4JK2200
205666	Refer for microbiological test	8HP2.00
210464	PENILE SWAB CULTURE NEGATIVE	L167DN
210515	HVS TRICHOMONAS VAGINALIS	L1670FT
212942	Sample culture	4J17.00
212962	Semen sent for C/S	4JL8.00
219515	SWAB CERVICAL ABNORMAL	L167FC
219570	HVS LACTOBACILLI	L1670FL
221698	Direct microscopy	31B1.00
222017	Sample: no organism isolated	4J11.00
222018	Sample: organism isolated	4J12.00
222020	Sample: bacteriology – general	4J2..00
222022	Sensitivity – bacteriology	4J2..13
222038	Microbiology NOS	4JZ..00
228578	MICROBIOLOGY REPORT ABNORMAL	L2MA
228611	HVS CULTURE NEGATIVE	L167FN
228613	SWAB CULTURE BACTERIAL GROWTH	L167XE
230862	Blood sent – infectious titres	43E1.00
231003	Parasite in urine	46H..15
231090	Microbiology	4J...00
231091	Sample – microbiological exam	4J1..00
231094	Sample: dir.micr.:no organism	4J71.00
231095	Bacteria on microscopy	4J72.11
231108	Urethral swab culture positive	4JK1000
231109	High vaginal swab: white cells seen	4JK2500
231110	Vaginal swab culture negative	4JK6.00
237538	MICROBIOLOGY REPORT	L2MR
237571	VAGINAL SWAB CULTURE POSITIVE	L167FZ
237574	SWAB CULTURE FUNGAL GROWTH	L167XC
237587	VIRAL TITRES	L189D
237617	HVS GARDNERELLA VAGINALIS	L1670FG
237618	HVS YEAST	L1670FY
240066	Sample: direct micr. organism	4J7..00
240075	High vaginal swab culture positive	4JK2000
240076	HVS culture – gardnerella vaginalis	4JK2300
240077	Low vaginal swab taken	4JK3.00
240078	Misc. sample for organism	4JL..00
246733	SWAB CERVICAL	L167FA
246735	URETHRAL SWAB CULTURE NEGATIVE	L167IN
249028	Swab sent to Lab	4147.00
249310	Culture – general	4J...11
249324	Cervical swab culture positive	4JK5000
258253	Blood – infect. titre negative	43E2.00
258485	Culture sensitivity	4J15.11
258486	Sample: microbiology NOS	4J1Z.00
258503	Urethral swab culture negative	4JK1100
258504	Vaginal swab culture positive	4JK7.00
258505	Penile swab culture positive	4JK8000
258506	Penile swab culture negative	4JK8100
265145	PENILE SWAB	L167D
265146	PENILE SWAB CULTURE POSITIVE	L167DP
265197	HVS WBC	L1670FW
267662	Urine microscopy: orgs/FB's	46H..00
267735	Sensitivity – microbiol.	4J...12
267736	Sample: organism sensitivity	4J15.00
267739	O/E: stained micr.: organism	4J8..00
267754	Vaginal swab taken	4JK..11
267755	Vulval swab taken	4JK4.00
267756	Penile swab taken	4JK8.00
267757	GUT swab NOS	4JKZ.00
274368	HVS EPITHELIAL CELLS	L1670FE
276782	Culture – bacteriology	4J2..12
276783	Sample sent for culture/sensit	4J22.00
276800	GUT sample taken for organism	4JK..00
276801	High vaginal swab taken	4JK2.00
276802	Cervical swab taken	4JK5.00
283373	HVS	L167F
283374	HVS CULTURE POSITIVE	L167FP
283375	VAGINAL SWAB CULTURE NEGATIVE	L167FY
285938	Microscopy, culture and sensitivities	4I16.00
285943	Sample: bacteria cultured	4J23.00
285955	Urethral swab taken	4JK1.00
285958	Microbiology test	4JQ..00
292462	MICROBIOLOGY REPORT NORMAL	L2MN
292509	SWAB CERVICAL NORMAL	L167FB
292511	URETHRAL SWAB CULTURE POSITIVE	L167IP
292515	SWAB CULTURE NO GROWTH	L167XB
295145	High vaginal swab: fungal organism isolated	4JK2400
295146	Cervical swab culture negative	4JK5100
297019	Microbiology report received	9ND3.00
301878	VAGINAL SWAB	L167FX
301879	URETHRAL SWAB	L167I
301882	SWAB CULTURE YEAST GROWTH	L167XD
308931	Bacterial antibody level	43e..00
309727	Microscopy	4JS..00
331709	Gram stain microscopy	4JS0.00
332043	Anaerobic culture	4J18.00
339918	Concentrate microscopy	4JS2.00
340342	Genital microscopy, culture and sensitivities	4I1 C.00
340745	Fluid microscopy, culture and sensitivities	4I1D.00
343815	Semen microscopy	49L..00
343816	Aerobic culture	4J19.00
344353	Additional urine tests	46h..00
345784	Culture for fungi	4J45.00
350883	Low vaginal swab taken by patient	4JKE.00
350959	Self taken low vaginal swab	4JKE.11

Appendix 2. Results of sensitivity analyses at the 42, 60, and 90-day management windows

	42 days, %	60 days, %	90 days, %
All cases in management analyses,	n = 3669	n = 3669	n = 3669
Managed in practice	53.7	50.7	46.8
No evidence of management	9.1	8.0	6.8
Evidence of care elsewhere	37.2	41.2	46.4
Of those managed in practice only	n = 1971	n = 1862	n = 1719
Any recommended drug	59.9	60.5	60.8
Any recommended drug + metronidazole	34.4	35.1	35.2
Chlamydia trachomatis test	57.0	57.9	59.0
Microbial test	52.2	53.8	55.5
Chlamydia trachomatis test and any treatment	36.3	37.1	37.6

Funding body

Access to the GPRD database was funded through the Medical Research Council's licence agreement with the Medicines and Healthcare products Regulatory Agency.

Ethics committee

The study was approved by the MHRA Independent Scientific Advisory Committee (protocol number 08_097).

Competing interests

The authors have stated that there are none.

Discuss this article

Contribute and read comments about this article on the Discussion Forum: http://www.rcgp.org.uk/bjgp-discuss