4.1 DIP payment reform for local patients has achieved some success, but still need further strengthening of regulation.
Evaluating the effectiveness of the DIP reform is a complex issue. The DIP payment method is similar to DRG, with the primary difference being the classification of disease groups. As such, it is worthwhile to look at the effectiveness of DIP reform using the same indicators used to evaluate the effectiveness of DRG. Theoretically, DIP should improve the efficiency of medical services and reduce inpatient costs and patients’ financial burdens. Choi’s study on the policy effects after South Korea transitioned from FFS to DRG payment revealed that implementing DRG payments significantly reduced patients’ overall medical expenses.[26] Similarly, Schuetza's research on the treatment quality of Swiss patients with community-acquired pneumonia found that there were no significant differences in clinical outcomes between the DRG and FFS, indicating a certain degree of moral hazard in the FFS method.[27] In Taiwan, employing DRG payments effectively reduced the length of hospital stays without compromising patient treatment quality.[28] Payment methods similar to DRG, such as DPC/PDPS in Japan, have also reduced patients’ medical burden and shortened the average length of hospital stay compared to FFS.[29] Research conducted by some Chinese scholars has likewise confirmed that DRG payment methods can produce cost-reduction and efficiency-enhancing effects compared to FFS method.[30–33]
However, it is essential to recognize that the effectiveness of the DIP payment method in reducing patients' medical burden is based on the premise that doctors tend to over-treat patients under the FFS method. For regions with stricter medical practice control and higher professional ethical awareness among doctors, over-treatment may be less prevalent even before the implementation of DIP payment methods. In such cases, the DIP payment method may not significantly reduce inpatient costs but rather guide hospitals to control their own costs by adopting the “retain surplus, share overruns” to achieve long-term stability of inpatient costs for each DIP group. As one of the cities with the most abundant medical resources and highest medical service management levels in China, the DIP payment method reform in Guangzhou has not shown an immediately or significantly decreasing trend in inpatient costs, as seen in Fig. 1. However, the changes in inpatient costs across different months have been relatively stable, indicating that the DIP payment method reform has achieved some of its intended effects for local patients.
But, as seen in Section 3.2, the average rate of cost belonging to the medical insurance catalog has significantly decreased after the policy implementation. This section further analyzes the changes in out-of-pocket expenses for local patients with the three diseases before and after the implementation of DIP payment. Table 6 shows the result. It can be seen that, firstly, for the I25.1 disease, the out-of-pocket expenses had a significantly increasing trend before policy implementation (P < 0.001). Although these expenses decreased by 885.69 yuan in the short term after the policy implementation (P < 0.000), the long-term trend remained unchanged (P < 0.000). After policy implementation, the monthly increase in out-of-pocket expenses remained at a significant rate of 50.12 yuan. Second, for the I63.9 and Z51.1 diseases, patients’ out-of-pocket expenses did not exhibit statistical significance in both the pre-implementation period and short-term effects of the policy, but both showed a significant long-term increasing trend (P < 0.000). Therefore, from the patient's perspective, their financial burden has not decreased with the implementation of DIP payment. The overall trend is still increasing, which means they have not enjoyed the benefits of the DIP payment method reform.
Table 6
The changes in the out-of-pocket expenses for local patient
Coefficient | I25.1 | I63.9 | Z51.1 |
Coef. | P | Coef. | P | Coef. | P |
β1 | 46.546 | 0.001 | -13.031 | 0.583 | -22.643 | 0.383 |
β2 | -885.690 | 0.000 | -204.379 | 0.237 | 27.925 | 0.901 |
β3 | 3.570 | 0.829 | 68.411 | 0.008 | 56.589 | 0.046 |
β0 | 5638.457 | 0.000 | 3180.307 | 0.000 | 2729.990 | 0.000 |
β8 | 50.116 | 0.000 | 55.381 | 0.000 | 32.588 | 0.001 |
We need to explain this phenomenon by focusing on the calculation method of hospitals’ income within the DIP payment in Guangzhou. Under the DIP payment framework in Guangzhou, the income from the DIP payment for a hospital can be represented by the following formula [34]:
$$\begin{array}{c}{x}_{i}={s}_{h} \times up\times {apr}_{h}\end{array}$$
Where \({x}_{i}\) represents a hospital’s medical insurance fund income under DIP payment, \({s}_{h}\) represents the sum score of all cases in the hospital, \(up\) represents the payment standard of unit score in DIP payment, and \({apr}_{h}\) represents the actual medical insurance reimbursement ratio calculated by FFS for the current year in this hospital.
Among them, \(up\) is calculated as follows:
$${cb}_{t}=mifb÷{apr}_{t}$$
$$\begin{array}{c}up = {cb}_{t}÷{s}_{t}\end{array}$$
The \({cb}_{t}\) represents the total DIP paid cost budget for the current year, which is obtained by dividing the \(mifb\) (representing the annual DIP payment medical insurance fund budget) with the \({apr}_{t}\) (representing the citywide actual medical insurance reimbursement ratio for the current year), and the \({s}_{t}\) represents the sum score of the citywide for all cases for the current year.
Combining the above equations with a series of simplifications, the following equation is finally obtained:
$$\begin{array}{c}{x}_{i}=\frac{{s}_{h}}{{s}_{t}}\times \frac{{apr}_{h}}{{apr}_{t}}\times mifb\end{array}$$
From this, it is clear that the income of the hospital is mainly influenced by \(\frac{{s}_{h}}{{s}_{t}}\) and \(\frac{{apr}_{h}}{{apr}_{t}}\). Relative to the \({s}_{h}\), hospitals can more easily achieve the regulation of \({apr}_{h}\). Under the assumption that the total hospital inpatient cost \({c}_{h}\), the total citywide inpatient cost \({c}_{t}\) and other conditions remain unchanged, and only ∆p reimbursable amounts are changed, the final income impact for a hospital can be calculated as
$${\varDelta x}_{i}=\frac{{s}_{h}}{{s}_{t}}\times mifb\times (\frac{\frac{{p}_{h}+\varDelta p}{{c}_{h}}}{\frac{{p}_{t}+\varDelta p}{{c}_{t}}}-\frac{\frac{{p}_{h}}{{c}_{h}}}{\frac{{p}_{t}}{{c}_{t}}})$$
$$=\frac{{s}_{h}}{{s}_{t}}\times \text{m}\text{i}\text{b}\times \left(\frac{{c}_{t}\varDelta p({p}_{t}-{p}_{h})}{{p}_{t}{c}_{h}({p}_{t}+\varDelta p)}\right)$$
Since\({c}_{h}\), \({c}_{t}\), \({p}_{h}\), \({p}_{t}\) are all constants greater than 0, and \({p}_{t}\ge {p}_{h}\), the relationship between \({\varDelta x}_{i}\) and \(\varDelta p\) can be simplified as
$${\varDelta x}_{i}\propto \frac{\varDelta p}{\varDelta p+{p}_{t}}$$
Taking practical factors into account, we know that \(\varDelta p\) is always less than \({p}_{t}\). As a result, the equation represents a continuously increasing function. When \(\varDelta p<0\), \({\varDelta x}_{i}\) is negative, which means a reduction in the income of hospitals, and the larger \(\left|\varDelta p\right|\), the more \({\varDelta x}_{i}\) decreases. When \(\varDelta p>0\), \({\varDelta x}_{i}\) is positive, indicating an increase in the income of hospitals, and the larger \(\varDelta p\), the more \({\varDelta x}_{i}\) increases. Therefore, theoretically, hospitals should aim to obtain more compensation from the health insurance fund by reducing the rate of patients’ out-of-pocket expenses and increasing the average rate of cost belonging to the medical insurance catalog. However, this contradicts the actual data results.
This may seem illogical, but when we look at the calculation rules for \({\varDelta x}_{i}\), firstly, the product of coefficients\(\frac{{s}_{h}}{{s}_{t}}\) and \(\frac{{c}_{t}({p}_{t}-{p}_{h})}{{p}_{t}{c}_{h}}\) is relatively stable, with certain constraints between them. This causes the variation of \(\frac{\varDelta p}{\varDelta p+{p}_{t}}\) not to be significantly scaled. Secondly, since \({p}_{t}\gg \left|\varDelta p\right|\), the final value of \(\frac{\varDelta p}{\varDelta p+{p}_{t}}\) is too small, and changes in \(\varDelta p\) within a certain range have minimal impact on the actual medical insurance fund income that hospitals can obtain.
Due to the characteristics of DIP payment, as long as patients’ inpatient costs are less than the product of the DIP group's score and the unit price, hospitals can get profit from the medical insurance fund. Thus, from the perspective of hospitals, it is appropriate to moderately increase patients’ out-of-pocket expenses and utilize more items outside the medical insurance catalog. On one hand, it has less impact on the amount of DIP medical insurance payment received at the end of the year.[35, 36] On the other hand, the out-of-pocket expenses are fully paid by patients upon discharge, easing the financial pressure on hospitals and facilitating the operation of their cash flow.
Therefore, for local patients who have experienced the DIP payment method reform, medical insurance management departments should place particular emphasis on assessing the proportion of patients' out-of-pocket expenses and further standardizing doctors’ clinical practices.
4.2 DIP payment reform for local patients fails to affect the other-insured-region patients at the same time, and there is still some over-treatment phenomenon
From the results in sections 3.3–3.4, the average inpatient costs of the three diseases for the other-insured-region patients are higher than those for local patients, particularly for I25.1 and I63.9. In addition, the average inpatient costs for the other-insured-region patients fluctuate more between months, we failed to find the impact of DIP payment reform for local patients on the other-insured-region patients. Is this due to the fact that the other-insured-region patients have relatively “more” severe conditions?[20] To investigate this, the paper further analyzed the total number of diagnoses and the proportion of patients who did not undergo surgery among the other-insured-region and local patients in 2018 and 2019. The results after the Mann-Whitney U test are shown in Table 6. It can be seen that the total number of diagnoses for the other-insured-region patients in both years was significantly lower than that for local patients. Relatively speaking, the proportion of the other-insured-region patients who did not receive surgical is higher. Only in 2019, the proportion of the other-insured-region patients with I25.1 disease who underwent surgery was higher than that of local patients.
Table 6
The difference in average numbers of diagnoses and the percentage of patients not receiving surgery in the two patient groups
Disease | Year | Average numbers of diagnoses | Percentage of patients not receiving surgery |
The other-insured-region patients | Local patients | P | The other-insured-region patients | Local patients |
I25.1 | 2018 | 4.65 | 7.42 | 0.000 | 47.05% | 44.56% |
2019 | 4.39 | 7.52 | 0.000 | 15.13% | 38.04% |
I63.9 | 2018 | 4.80 | 5.97 | 0.000 | 89.05% | 80.19% |
2019 | 5.15 | 6.53 | 0.000 | 63.89% | 33.51% |
Z51.1 | 2018 | 1.90 | 5.28 | 0.000 | 85.40% | 53.57% |
2019 | 2.50 | 5.45 | 0.000 | 71.18% | 39.29% |
Considering the number and difficulty level of surgeries performed on patients who received surgical (see Table 7), on one hand, there is a certain pattern in the comparison of the number of surgeries performed on the other-insured-region and local patients for the three diseases over the two years. Specifically, the other-insured-region patients with I25.1 and Z51.1 diseases had a significantly higher total number of surgeries per person compared to local patients, while there was no significant difference in the total number of surgeries performed on I63.9 patients between the two groups. On the other hand, except for the Z51.1 group in 2019, the proportion of level 3 and 4 surgeries performed on the other-insured-region patients was higher than that for local patients in the same period, especially for I25.1 and I63.9 groups, where the number of level 3 and 4 surgeries performed on the other-insured-region patients was higher than that for local patients.
Table 7
The difference of the number and difficulty level of surgeries in the two patient groups
Disease | Year | Average number of surgeries | Proportion of level 3 and 4 surgeries performed | Average number of level 3 and 4 surgery per case |
The other-insured-region patients | Local patients | P | The other-insured-region patients | Local patients | The other-insured-region patients | Local patients |
I25.1 | 2018 | 3.23 | 3.04 | 0.004 | 50.88% | 47.30% | 3.68 | 3.33 |
2019 | 3.59 | 3.18 | 0.000 | 60.76% | 50.14% | 4.07 | 3.39 |
I63.9 | 2018 | 2.10 | 2.09 | 0.671 | 17.21% | 14.69% | 2.28 | 2.1 |
2019 | 2.16 | 1.98 | 0.654 | 21.90% | 9.92% | 2.64 | 2.28 |
Z51.1 | 2018 | 1.55 | 1.38 | 0.000 | 8.35% | 7.13% | 1.28 | 1.37 |
2019 | 1.43 | 1.42 | 0.000 | 3.59% | 5.64% | 1.19 | 1.29 |
These two sets of data can better explain the following issues: firstly, they objectively demonstrate the characteristics of the other-insured-region patients, i.e. compared with local patients, the other-insured-region patients have a greater degree of disease differentiation, which causes a greater fluctuation of the mean value of off-site patients by month in Fig. 3; secondly, relatively speaking, for patients who undergo surgical treatment, the other-insured-region patients receive a higher number of surgeries and face more complex surgeries, indicating more severe disease progression than local patients.
However, this does not mean that the relatively higher costs for the other-insured-region patients are completely justified. The average inpatient costs of the other-insured-region and local patients for the three types of diseases in 2018 and 2019, classified by the highest level of surgery performed, are shown in Fig. 5.
As can be seen, firstly, the other-insured-region patients have higher average inpatient costs than local patients in all three diseases, but the difference in costs between the two types of patients varies among different diseases. I25.1 and I63.9 diseases have a more considerable difference in average inpatient costs between the other-insured-region and local patients, while the difference between Z51.1 patient groups is relatively more minor, which is consistent with the phenomenon shown in Fig. 1, wherein the I25.1 and I63.9 diseases, the other-insured-region patients have significantly higher average inpatient costs than local patients each month, while in the Z51.1 disease is slightly higher than local patients. Secondly, it can be observed that the inpatient costs of local patients increase gradually with the difficulty of the surgery they undergo, while for the other-insured-region patients, especially those in I25.1 and I63.9 diseases, the average inpatient costs of patients without surgery are higher than those of patients who underwent lower-level surgeries and are significantly higher than local patients who did not receive surgical treatment, accounting for 2.56 times and 1.89 times of the corresponding reference, respectively.
Moreover, as shown in Fig. 4, before the implementation of the DIP payment reform, the average rate of cost belonging to the medical insurance catalog for the other-insured-region patients of the three diseases was significantly lower than those of local patients, indicating that doctors prescribed more services outside the insurance scope for the other-insured-region patients. Notably, after the implementation of the DIP payment reform for local patients, the other-insured-region patients still retain the FFS payment method, and hospitals may transfer the costs of the payment method reform to the other-insured-region patients.[13, 37, 38] In China, for some high-level hospitals, the scale of the other-insured-region patients has exceeded that of local patients, and the source of income of the hospital's medical insurance fund has gradually changed from local to nationwide, which makes the medical insurance payment reform for the other-insured-region patients an urgent issue.
By building a DRG/DIP payment system for the other-insured-region patients with the “same disease, same group, same score, same treatment, and same price” as local patients, we can achieve the purpose of guiding doctors’ behavior by using medical insurance payment policy, and thus reduce the moral risk of over-care by doctors. We can realize the use of medical insurance payment policies to restrain doctors’ behavior, thus reducing the moral risk of over-treatment by doctors.