Research and analysis

Cervical screening: invasive cervical cancer audit 2016 to 2019: Appendix A: data completeness

Published 30 November 2023

Applies to England

1. Data Completeness and limitations

When considering the findings presented in this report, the varying degree of completeness of the available information should be considered. The difficulties involved in ensuring the completeness of essential data fields are described below.

It is rare for data to be reported as missing, but missing data should be distinguished from incompleteness of record.  Missing data may be unavailable (e.g. where a death certificate, which does not provide information about cancer staging, has been used), or may not yet have been recorded as part of the audit. For this reason, we have used the term ‘none recorded’ to describe cases where final stage is still pending and the term ‘none available’ to describe cases where after considerable effort, no staging data has been available.

Other cases may be subject to reporting delays, having been submitted to the audit before all essential fields could be completed. In these instances, missing fields are updated as and when data become available, with the result that complete information may not be received for some months after the case has been registered. An additional challenge, which can create further delay, is the need to coordinate between the various aspects of the audit process when a case of cervical cancer is diagnosed.

2. Cancers and population controls

Cases of cervical cancer are identified by NHS hospital staff (primarily via gynae-oncology), and confirmed by histology. A small proportion of cancers will be missed by the audit, and a very small proportion will be excluded because the patients are not registered with an NHS GP. Table 1 illustrates the limited extent of this problem, comparing the number of registrations for cervical cancer in a given calendar year with the number of cases picked up by the audit over the same period of time.

Controls are selected randomly (subject to matching) from people registered with an NHS GP. All those selected are included in the audit.

3. Dealing with missing values

Cases reported in the MB1 series (Cancer Registration Statistics in England, Office for National Statistics) between 2016 and 2018 were compared to those recorded in the audit for the same period, by age at diagnosis (Table 1). The aim was to ascertain whether there is a subset of cases for which a delay in their inclusion in the audit is more likely, and whether this is related to age at diagnosis. 81% of the cervical cancer cases registered in England between 2016 and 2018 were recorded in the audit for the same period. However, the audit data are more likely to include cases diagnosed in people between the ages of 25 and 64 (audit includes 85% of all registered cancers in this age group), than cases diagnosed in people over the age of 65 (audit includes 64% of registered cancers in this age group). The completeness of the audit data, compared with MB1 decreased with increasing age at diagnosis.

We assessed the completeness of audit data for FIGO stage by comparing the distribution of staged cancers diagnosed netween April  2011 and March 2012 across 5 audit years (Table 2). The table shows that if we were to assign a stage to cases with this information missing, assuming that stage was missing at random, we would be overestimating the proportion of cases diagnosed with early stage cancer. For example, based on data received as of October 2011 we would have assumed 47.6% of cases diagnosed between April 2011 and March 2012 had stage IA cancer. However, by October 2012 that proportion decreased to 42% and remained as such since then. This suggests that cases with unknown FIGO stage are more likely to be advanced stage cancer.

In the audit reports published in July 2011 and May 2012, we assumed that data for FIGO staging was missing at random, which would have led to an overestimation of the proportion of stage IA cancers and an underestimation of the proportion of stage II+ cancers. For this report (and the 2 previous reports), we have used a more complicated model that takes into account the differential delays in obtaining stage.[1]

A- 1a  National Cancer Registration Statistics (MB1 Series), published by the Office for National Statistics, compared with those reported in the audit

	Age at diagnosis
Total cases reported
<20	20-24	25-29	30-34	35-39	40-44	45-49	50-55	55-59	60-64	65-69	70-74	75-79	80-84	85+	Total	25-64
MB1 series* 2016/18   (n)	5	156	966	1093	936	826	802	595	563	473	348	349	259	246	236	7853	6254
Audit 2016/18   (n)	1	129	856	967	824	727	655	483	458	352	248	230	176	133	130	6369	5322
Difference (n)	4	27	110	126	112	99	147	112	105	121	100	119	83	113	106	1484	932
Proportion (%)	20.0	82.7	88.6	88.5	88.0	88.0	81.7	81.2	81.3	74.4	71.3	65.9	68.0	54.1	55.1	81.1	85.1

• MB1 Cancer Statistics are published by calendar year, audit data are normally reported by financial year (1 Apr to 31 Mar).

A -1b Cancers in people aged 25 to 64, diagnosed between April 2011 and March 2012

	Observed stage by year of Audit data                                                   (n)							Proportion assuming missing at random (%)
Received as of	IA	IB	II	III+	IB+	None recorded	Total	IA	IB	II+
Oct-11	285	207	50	28	29	124	723	47.6	38.1	14.3
Oct-12	656	520	169	124	73	161	1703	42.5	36.8	20.7
Oct-13	691	562	194	152	73	135	1807	41.3	36.3	22.4
Oct-17	774	601	214	172	71	117	1949	42.2	35.2	22.6
Oct-19	773	600	209	171	77	115	1945	42.2	35.4	22.4

A-2a Proportion of essential data collected for cases in section A: personal and cancer details

		Section A: Essential fields
		Date of Birth	Date of Diagnosis	Stage*	Histology
	Cases	n	%	n	%	n	%	n	%
Current report (2016/17-2018/19)	6,369	6,369	100	6,369	100	6,200	97.3	5617	88.2
Fifth report (2013/14-2015/16)	6,028	6,028	100	6,028	100	5,718	94.9	5,490	91.1
Fourth report (2009/10-2012/13)	8,784	8,784	100	8,784	100	8,014	91.2	8,543	97.3
Third report (2009/10-2011/12)	6,508	6,508	100	6,508	100	5,901	90.7	6,336	97.4
Second report (2007/08-2010/11)	8,566	8,566	100	8,566	100	7,423	86.7	8,197	95.7
First report (2007/08-2009/10)	6,231	6,231	100	6,231	100	5,197	83.4	5,922	95.0

*Cases where data collection is complete and stage is missing are considered to be staged as a reasonable amount of effort has been made to collect the data. Incomplete cases with a stage recorded as X (or missing) are considered not to have stage. Please refer to section 6 for full details regarding missing data.

A-2b Proportion of data collected for cases in section A: personal and cancer details

		Treatment (in those with known tx, excluding those reported as none*)		Treatment (in those with tx recorded including those recorded as none)		Index of Multiple Deprivation                     (Cases)			Index of Multiple Deprivation                  (Controls)
	Cases	n	%	n	%	n	%	Controls	n	%
Current report (2016/17-2018/19)	6,369	4,766	74.8	4,926	77.3	5,667	89.0	12,392	8,782	70.9
Fifth report (2013/14-2015/16)	6,028	3,990	66.2	4,254	70.6	5,224	86.7	11,580	8,451	73.0
Fourth report (2009/10-2012/13)	8,784	5,970	68.0	6,183	70.4	6,843	77.9	17,270	7,345	42.5
Third report (2009/10-2011/12)	6,508	4,146	63.7	4,394	67.5	5,104	78.4	12,841	4,423	34.4
Second report (2007/08-2010/11)	8,566	5,199	60.7	5,675	66.3	6,485	75.7	16,920	7,964	47.1
First report (2007/08-2009/10)	6,231	3,086	49.5	3,382	54.3	4,723	75.8	12,335	5,947	48.2

*Where treatment was recorded as “None” we assume it means “none other than palliative care”. Attempts have been made to clarify this issue and there is now a category for palliative care; however some misclassification may still remain and therefore they are excluded from this column

A-3 Proportion of cases with FIGO stage reported as none recorded, none available* or IB or worse (1B+), by age and audit year ( from April 2016 to March 2019)

	None recorded	None available	IB+ (NOS)	Total
Age	%	%	%	%
<25	0.0	0.0	4.0	4.0
25-49	1.9	1.0	0.8	3.6
50-64	3.9	2.6	1.9	8.3
65+	4.3	3.4	1.7	9.4
Audit Year
2016/17	1.9	1.8	1.3	5.0
2017/18	2.2	1.6	0.8	4.6
2018/19	3.9	1.5	1.4	6.8
Previous reports
Current report (2016/17-2018/19)	2.7	1.6	1.2	5.4
Fifth report (2013/14-2015/16)	2.2	5.1	2.2	9.5
Fourth report (2009/10-2012/13)	8.8	1.9	3.8	14.5
Third report (2009/10-2011/12)	9.3	1.1	4.4	14.8
Second report (2007/08-2010/11)	12.0	1.6	4.2	17.8
First report (2007/08-2009/10)	16.6	N/A	4.4	21.0

• Where stage is reported as none available instead of none recorded a reasonable amount of effort has been made to find the stage, but none has been available. This is derived from cases recorded as “Audit complete” which means that no further details are being sought for these people. The option to report cases as “none available” has only been available to all SQAs since April 2012

A-4 Proportion of data collected for cases in section B: cytology

Section B: Cytology
			Completeness of data among recorded cytology tests
			Date test was taken		Result of Testb		Action Codeb
Audit year	Cases	Tests on all casesa	n	%	n	%	n	%
Current report (2016/17-2018/19)	6,369	24,334	24,334	100	24,210	99.5	24,312	99.9
Fifth report (2013/14-2015/16)	6,028	21,764	21,764	100	21,707	99.7	21,746	99.9
Fourth report (2009/10-2012/13)	8,784	35,810	35,810	100	35,803	100	35,781	99.9
Third report (2009/10-2011/12)	6,508	26,619	26,619	100	26,619	100	26,594	99.9
Second report (2007/08-2010/11)	8,567	34,910	34,910	100	34,910	100	34,870	99.9
First report (2007/08-2009/10)	6,231	25,972	25,972	100	25,954	100	25,951	99.9

a. Cytology tests known to the Audit and taken before diagnosis

b. Cytology data obtained directly from Open Exeter should have all three data fields complete. Missing data, we believe, is the result of inclusion into the audit of cytology tests taken before the programme started in 1988 and a few slides that were found in the laboratory, but not recorded on Exeter. These tests will not have “Action Code” as this field is generated by Exeter.

A-5 Proportion of data collected for cases in section C: colposcopy

	Section C: Colposcopy
Audit report	Cases with an Action Code of suspend	Cases with a suspend and a colposcopy		Cases with a colp but no suspend	No. of Colp appts	Date of colp		Satisfactory exam or DNA*		Colp procedure
		n	%	n	n	n	%	n	%	n
Current report (2016/17-2018/19)	3,893	2944	75.6	772	5,987	5987	100	4934	82	4409
Fifth report (2013/14-2015/16)	3,674	2,397	65.2	627	4,378	4378	100	4,378	100	3,601
Fourth report (2009/10-2012/13)	6,073	3,963	65.3	494	6,823	6823	100	6,823	100	5,479
Third report (2009/10-2011/12)	4,523	2,843	62.9	430	5,195	5195	100	4,347	84	4,287
Second report (2007/08-2010/11)	5,884	3,604	61.3	647	7,167	7,167	100	5,942	83	5,620
First report (2007/08-2009/10)	4,308	2,412	56.0	557	4,348	4,348	100	3,445	79	3,249

• DNA, did not attend