A national study  By Dr S Williams
 

Abstract

Objectives To describe the population of NHS employees retired on grounds of ill-health and the work-relatedness of their health problem; to estimate annual ill-health retirement (IHR) rates for England and Wales and the direct costs to the NHS Pensions Agency.

Design Cross sectional survey using data from IHR application forms plus postal questionnaires.

Subjects 1,994 NHS staff awarded ill-health retirement by the NHS Pensions Agency.

Main outcome measures Diagnosis leading to IHR and its work-relatedness, annual IHR rates, quality of life at the time of retirement and additional direct costs of IHR .

Results Data from application forms were obtained on 1,994 retirees whose mean age was 51.6 years. The main diagnostic categories were musculo-skeletal (50%), psychiatric (19%) and cardiovascular (11%). Annual IHR rate was 5.5 per 1,000 employees. The estimated additional cost to the NHS pensions agency for 1998/9 retirees up to age 70 is £416 million.

Questionnaires were returned by 1,317 (66%) retirees, with a mean length of NHS employment of 21.2 years. A third of respondents believed their ill-health was caused through work. The SF36 physical and mental component scores (PCS 30.8; MCS 38.7) were significantly lower compared with the UK general population and the US population with chronic disease.

Conclusions Ill-Health retirement represents a major burden on the human and financial resources of the NHS. A third of ill-health leading to early retirement may be preventable with workplace interventions. Redeployment should also be considered as an alternative to IHR with the current recruitment and retention crisis in the NHS.

Introduction

Approximately one million people are employed by the National Health Service (NHS) in England and Wales. The vast majority (96%) belong to the NHS occupational pension scheme¹. The scheme allows for early retirement due to ill-health, provided the member has paid at least two years of contributions and is "incapable of discharging efficiently the duties of their employment by reasons of permanent ill-health or infirmity of mind or body", where permanent means until the applicant’s normal age of retirement². There is no requirement for the employer to attempt redeployment to a more suitable job taking into account the member’s health problem. Very little is known about staff granted ill-health retirement (IHR). 5,469 staff in England and Wales were granted IHR in 1998/9¹. As far as we are aware, the total cost of IHR has not previously been reported.

We hypothesise that a proportion of health problems for which NHS staff are retired are caused through work and are potentially preventable. Avoiding these early retirements would reduce the financial burden on the pension scheme and employer. This study was designed to describe the population of NHS employees retired on grounds of ill-health and the work-relatedness of their health problem. We estimate annual IHR rates for England and Wales and the direct costs to the NHS Pensions Agency.

Methods

Sampling and data collection

The NHS Pensions Agency allowed us access to 1,994 ill-health retirement forms of consecutive successful applicants from England and Wales who were granted IHR between April-August 1998. The following information was extracted for each retiree: age, sex, job title, sickness absence in the preceding year, medical reason for IHR and the manager’s response to the question about incapacity being due to employment.

We then wrote to all retirees to obtain information about type of employment (part-time or full-time), duration of NHS service and their belief about whether their ill-health was caused by work. Additionally, we asked them to complete the Short Form 36 (SF36) questionnaire, an internationally recognised and validated measure of health-related quality of life³. This contains 36 items which measure eight dimensions on a 0-100 scale: physical functioning, social functioning, role limitations due to physical / emotional problems, mental health, vitality, bodily pain and general health perception. These are used to calculate the physical (PCS) and mental component scores (MCS), each having a general population norm of 50⁴. Higher scores indicate better quality of life.

Annual number of ill-health retirees in each occupational group was estimated for England and Wales by applying our frequency distributions to the total number of retirees reported by the Pensions Agency for the financial year 1998/9¹. National statistical bulletins provided numbers for the NHS workforce in England and Wales as at September 1997^5,6,7. These were used as denominator data in calculating annual IHR rates, overall and for each occupation.

Quality of life in the subsample of retirees residing in England was compared with that reported from a national survey (Health Survey for England 1996⁸). Comparative SF-36 scores and data on potentially confounding sociodemographic factors were obtained for 10,348 participants in the national survey aged 25-64 years (corresponding to the IHR sample age limits).

Costs to the Pensions Agency

We estimated the additional sum paid to the 5,469 staff retired in 1998/9, from retirement to age 70. This sum consists of enhancements to the lump sum paid at retirement, the pension paid before the normal age of retirement (60 years) and the pension enhancements from age 60 to 70 (Annex).

Statistical Analysis

Chi-squared and unpaired t-tests were used to test differences in proportions and means. Non-parametric methods (Kruskal Wallis tests) were used for variables with skewed distributions. Adjusted differences in quality of life between retirees and the general population were obtained from multiple regression models using PCS and MCS scores as outcome variables and controlling for confounding factors (sex, age, social class and ethnicity). Confidence intervals were calculated with 95% probability. All statistical tests were two-sided and used a 0.05 level of significance. Statistical analyses were performed using SPSS 9.0 for Windows.

Results

The whole sample of 1,994 ill-health retirees

Sociodemographic characteristics. The mean age (standard deviation) at retirement was 51.6 years (SD, 7.4). The age by sex distribution of our sample is shown in Table 1. Sixty-nine percent of our sample were aged 50 and over. The 72 retirees aged less than 35 years extrapolates to an annual figure of more than 200 young employees leaving the NHS on grounds of ill-health.

Diagnostic categories. The commonest reason for IHR was musculo-skeletal conditions (961) followed by psychiatric (376) and cardiovascular conditions (208). The remaining 399 cases were caused by neurological, respiratory, endocrine, malignant and other conditions. Table 2 presents the relative frequency of diagnostic categories for each occupational group. Inverse linear trends (p<0.001) can be observed in the frequency of musculo-skeletal and psychiatric diagnoses, the former being relatively more common in jobs with a higher manual element.

Ill-health retirement rates. Estimated annual number of retirees in each occupational group, denominator data and IHR rates are presented in Table 3. Although the overall annual rate of ill-health retirement is 5.5 per 1000 NHS employees, much higher rates are seen in ambulance workers and health care assistants/support staff.

Costs to the Pensions Agency. We estimate that NHS staff retired in 1998/99 will cost the Pensions Agency an additional £416 million between retirement and age 70. The enhancement to the lump sum paid at the time of IHR was £23 million, £282 million in pension will be paid up to age 60 while the remaining £111 million represents pension enhancements after age 60. Two thirds of the total sum will be received by nurses (43%) and doctors (25%).

Questionnaire respondents

Bias analyses. Of the 1,994 retirees, 1,317 (66%) agreed to participate in the study and returned completed questionnaires. 461 (23%) declined to take part and 216 (11%) did not reply. The three significant differences between participants and non-participants were that non-participants were one year younger, had one month longer sickness absence in the year prior to retirement and had been retired more frequently for psychiatric and less for musculo-skeletal problems. There were no significant differences in terms of sex, social class, occupational group, type of employment (full-time or part-time) and manager’s opinion on work-relatedness of ill-health.

Employment status. Sixty-nine percent of respondents had retired from full-time as opposed to part-time employment. The mean number of years worked in the NHS was 21.2 (SD 9.1, range 2-43).

Perception of work-relatedness of ill-health. The retirees’ and managers’ perceptions of the work-relatedness of ill-health were compared in a subgroup of 1,008 participants for whom this information was available (Table 4). While a third of retirees believed their ill-health was caused by work, this was the case more frequently for musculo-skeletal and psychiatric conditions. The manager agreed that ill-health was due to employment in less than half of these cases.

Quality of life. The figure shows SF36 scores for ill health retirees living in England compared with the general population of England. The retirees reported lower health related quality of life than the general population in terms of physical and mental component scores (30.8 vs. 50.4 and 38.7 vs. 51.1, respectively). These unadjusted differences resulted from lower scores on all eight individual scales, in particular the scale measuring role limitations due to physical problems.

Adjusted differences calculated from multiple regression models were consistent with the unadjusted findings. Physical and mental component scores were 17.5 (95% CI; 16.8 to 18.1) and 12.2 (95% CI; 11.5 to 12.9) points lower in our sample compared to the general population after controlling for the confounding effect of sex, age, social class and ethnicity.

Discussion

This is the first major study of ill-health retirement in the NHS. Access to the IHR application forms allowed us to obtain detailed information on the whole sample. We achieved a good response rate for the postal questionnaire survey, given the sensitive nature of our subject. On most characteristics those who returned the questionnaires were similar to the whole sample of ill-health retirees.

The mean age and length of NHS service indicate that, on average, retirees lost more than 8 years of working life while their employers had to replace very experienced staff. The most frequent conditions for which our sample were awarded IHR were musculo-skeletal and psychiatric. This was also found in a recent study of pension schemes in six UK industries⁹. The distribution of diagnoses by occupational groups shows a trend of increasing musculo-skeletal problems as jobs become more manual. This may be due to a high incidence of work-related injuries in these jobs. Alternatively, musculo-skeletal conditions leading to incapacity may preclude return to work with a large manual component.

The high rates of IHR seen in health care assistants/support staff and ambulance workers may also be explained by the large manual component to their work. Overall the largest group of retirees was nurses and midwives, reflecting their large numbers in the NHS.

We found a large discrepancy between the managers’ and retirees’ views on the work-relatedness of ill-health leading to IHR. It may be that managers are unaware of the link, or reluctant to acknowledge work-relatedness of ill-health fearing the consequences of admitting liability. There was greater agreement for musculo-skeletal causes than for psychiatric. This is not surprising as many musculo-skeletal injuries are caused by a specific accident at work which is witnessed by colleagues. Psychiatric illness, on the other hand, is probably more multi-factorial and the work-related component may be a combination of workload pressures, lack of support and conflict with individual colleagues^10,11. Such factors are less tangible and it may be harder for the worker to gain recognition by their employer of the work-relatedness of their illness.

Our cost estimates are based on average figures from our sample and on mid-range salaries assumed for each occupational group. According to our assumptions we estimated that the Pensions Agency will pay £416 million over the next 18 years, in addition to the normal pension, for 1998/9 retirees. This could be reduced by up to a third (£139 million) if the NHS were able to prevent ill-health and injury caused through work. A similar figure was obtained by the National Audit Office who looked at the cost of IHR due only to work-related illness or injury. They estimated that staff retired in 1994/5 will receive £150 million over a period of 25 years in addition to the normal pension¹².

A further reduction in the IHR costs to the Pensions Agency could be achieved through redeployment of those retirees whose health problems were not caused through work but prevent them continuing in their normal job. For example a nurse with musculoskeletal problems could be retrained in IT skills and redeployed to a more sedentary job.

Our sample reported poorer quality of life at the time of retirement compared with a random sample of the UK general population. After adjustment for differences in sociodemographic characteristics, physical and mental quality of life scores remained significantly lower in our sample. The scores were also lower compared with similar chronic disease groups in the US general population⁴. For example, the physical component score in those retired for musculoskeletal problems (26.6) was much lower than the score reported in the US by people with backpain/sciatica or arthritis (43.1). Similarly, the mental component score in our retires was 25.2 compared with 34.8 in the US population with clinical depression. These comparisons support the use of the SF-36 questionnaire as a sensitive measure of quality of life among ill-health retirees, and could also imply that the decision to grant ill-health retirement in the UK health service is appropriately made.

This study measured the impact of ill-health retirement in terms of loss of working years, loss of expertise and costs to the Pensions Agency. Our findings suggest that up to a third of IHR may have been caused by work, indicating the need for preventative workplace interventions. There is a need for high quality research to identify effective interventions. For some retirees, redeployment may be an alternative, allowing the NHS to keep experienced staff at a time of crisis in recruitment and retention. We plan to contact our cohort at intervals to find out whether they have found alternative work and, if so, to identify predictors of re-employment. SF-36 questionnaires will provide further insight into their quality of life after retirement.

ANNEX

Assumptions used in estimating direct costs of ill-health retirement to the NHS Pensions Agency

• The mean age and length of NHS employment in our sample were representative of all NHS staff retired in 1998/9
• IHR pension will be paid from age of retirement up to 60 years, followed by normal pension paid for a further 10-year period (cut-off point arbitrarily set at 70 years of age)
• all retirees were considered to have made pension contributions for the full duration of their employment in the NHS
• lump sum and pension enhancements were calculated for 6²/₃ years (the standard used by the Pensions Agency) in addition to employee’s actual length of NHS service
• a mid-range annual salary at ill-health retirement was assumed for each occupational group: doctors/surgeons, £60K; nurses/midwives, £17.5K; technical/professional staff, £22K; administration/estates, £15K; healthcare assistants, £10K; ancillary staff, £8K; ambulance workers, £10K
• final costs due to IHR are made up of lump sum enhancements, early pension payments (before 60 years), and enhancements to the final pension (after 60 years)
• future costs were adjusted for an annual inflation rate of 3%.

References

1. The NHS Pensions Agency Annual Report and Accounts 1998-99. London: The Stationery Office, 1999.
2. National Health Service Pension Scheme Regulations 1995 (Regulation E2). Statutory Instrument 300. London: The Stationery Office, 1995.
3. Ware JE Jr, Snow KK, Kosinski M, Gandek B. SF-36 Health Survey: Manual and interpretation guide. Boston: The Health Institute, New England Medical Center, 1993.
4. Ware JE Jr, Kosinski MA, Keller SD. SF-36 physical and mental component summary measures: A user's manual. Boston: The Health Institute, New England Medical Center, 1994.
5. The Government Statistical Service. Health and Personal Social Services Statistics for England. London: The Stationery Office, 1998.
6. Directly employed NHS non-medical and dental staff in post at 30^th September 1997. Cardiff: The National Assembly for Wales Statistical Directorate, 1998.
7. Directly employed NHS medical staff in post at 30^th September 1997. Cardiff: The National Assembly for Wales Statistical Directorate, 1998.
8. Health Survey for England, 1996 [computer file], 2nd ed. Joint Health Surveys Unit of Social and Community Planning Research and University College London, Colchester, Essex: The Data Archive (distributor), 1999 Nov. SN: 3886.
9. Poole CJM. Retirement on grounds of ill-health: cross sectional survey in six organisations in United Kingdom. BMJ 1997;314:929-32.
10. Karasek Jr RA. Job demands, job decision latitude, and mental strain: implications for job redesign. Administrative Science Quarterly 1979;24:285-311.
11. Williams S, Michie S, Pattani S. Improving the Health of the NHS workforce; report of the Partnership on the health of the NHS workforce. The Nuffield Trust, 1998.
12. Health and Safety in NHS Acute Hospital Trusts in England. Report by the Comptroller and Auditor General. National Audit Office, 1996.

Acknowledgements

We thank Richard Rance for assisting with questionnaire design, data collection and input. We thank the NHS Pensions Agency for financial support and for allowing us access to their records. We thank Medical and Industrial Services Ltd for their help in mailing the questionnaires and retrieving data from the ill-health retirement application forms. We also thank Prof. Charles Normand and Mr Gary Cole for advice on cost calculations.

Contributors

SP, NC and SW designed the study, devised the data collection instruments and wrote the paper. SP and SW obtained the funding. SP oversaw data collection and obtained ethical approval. NC analysed the data. SW conceived the idea and oversaw the study. SP is guarantor of the study.

Funding

Part funded by the NHS Pensions Agency

Competing interests

None

Ethics approval

Given by the Royal Free Hampstead NHS Trust Ethics Committee.

Disclaimer for the use of Health Survey for England 1996 data

The copyright holder, the original data producer, the relevant funding agencies and The Data Archive bear no responsibility for the further data analysis or interpretation in this paper.

Table 1. Age-sex distribution in the whole sample (numbers in parentheses are percentages)
 

Table 2. Type of medical condition leading to ill health retirement by occupation (numbers in parentheses are percentages)
 

^{1 Retirees in the ‘Other’ occupational category were excluded n=50}

Table 3. Estimated annual ill health retirement rates in six occupational groups in England and Wales
 

^{1 Obtained by pooling figures reported separately for England and Wales [5,6,7]}

^{2 Retirees in the ‘Other’ occupational category were excluded n=50}

^{3 The number of retirees in each occupational group estimated by applying the IHR sample distribution to the total number of staff granted IHR in 1998/9 (N=5,469)}

Table 4. Agreement between retirees and their managers on work-relatedness of medical condition leading to IHR ¹ (numbers in parentheses are percentages)
 

^{1 The retiree’s and their manager’s views were both available on a subsample of 1,008 records}

^{2 Expressed as percentage of retirees whose managers agreed on work-relatedness of medical condition}

Figure

^{1 Health Survey for England (N=10,384); 2 Sub-sample of residents of England (n=1,063);}